Bulletin of Mathematical Biology (2010) 72: 1867–1891

DOI 10.1007/s11538-010-9514-7

O R I G I N A L A RT I C L E

Mathematical Model of Hyperbaric Oxygen Therapy

Applied to Chronic Diabetic Wounds

Jennifer A. Flegga,∗ , Helen M. Byrneb , D.L. Sean McElwaina

a
Discipline of Mathematical Sciences and Institute of Health and Biomedical Innovation,
Queensland University of Technology, GPO Box 2434, Brisbane, Queensland 4001,
Australia
b
School of Mathematical Sciences, University of Nottingham, Nottingham NG7 2RD, UK

Received: 11 February 2009 / Accepted: 29 January 2010 / Published online: 5 March 2010
© Society for Mathematical Biology 2010

Abstract The failure of certain wounds to heal (including diabetic foot ulcers) is a sig-
nificant socioeconomic issue for countries worldwide. There is much debate about the
best way to treat these wounds and one approach that is shrouded with controversy is
hyperbaric oxygen therapy (HBOT), a technique that can reduce the risk of amputation in
diabetic patients.
In this paper, we develop a six species mathematical model of wound healing angio-
genesis and use it to investigate the effectiveness of HBOT, compare the response to dif-
ferent HBOT protocols and study the effect of HBOT on the healing of diabetic wounds
that fail to heal for a variety of reasons. We vary the pressure level (1 atm–3 atm), per-
centage of oxygen inspired by the patient (21%–100%), session duration (0–180 minutes)
and frequency (twice per day–once per week) and compare the simulated wound areas
associated with different protocols after three weeks of treatment.
We consider a variety of etiologies of wound chronicity and show that HBOT is only
effective in treating certain causes of chronic wounds. For a wound that fails to heal due
to excessive, oxygen-consuming bacteria, we show that intermittent HBOT can accel-
erate the healing of a chronic wound but that sessions should be continued until com-
plete healing is observed. Importantly, we also demonstrate that normobaric oxygen is
not a replacement for HBOT and supernormal healing is not an expected outcome. Our
simulations illustrate that HBOT has little benefit for treating normal wounds, and that
exposing a patient to fewer, longer sessions of oxygen is not an appropriate treatment
option.

Keywords Hyperbaric oxygen · Optimal protocol · Chronic wound · Mathematical

modelling · Diabetes

∗ Corresponding author.
E-mail addresses: j.flegg@qut.edu.au (Jennifer A. Flegg), Helen.Byrne@nottingham.ac.uk (Helen
M. Byrne), s.mcelwain@qut.edu.au (D.L. Sean McElwain).
1868 Flegg et al.

Abbreviations
HBOT hyperbaric oxygen therapy
ECM extracellular matrix
EC endothelial cells
PDGF platelet derived growth factor
VEGF vascular endothelial growth factor
atm atmospheres
MMPs matrix metalloproteinases
PDEs partial differential equations

1. Introduction

A chronic wound is one that does not progress through the stages of wound healing that
normally lead to a successful outcome. Often chronic wounds are a surface manifestation
of an underlying disease such as diabetes mellitus, arterial disease, and vascular insuf-
ficiency (Mathieu, 2002). The treatment of such wounds is expensive since the patients
often require ongoing regular care or hospitalisation. Among people with diabetes, ap-
proximately 15% will suffer from a foot ulcer and up to 24% of those with a foot ulcer
will require amputation of a limb (Anonymous, 1999). A recent estimate suggests that
it costs the US health care system up to US$25 billion annually to treat patients with
nonhealing wounds (Sen et al., 2009). Grey and Harding (2006) have recently edited an
excellent series of papers that provide a comprehensive review of wound healing and cur-
rent standard treatments.

1.1. Wound healing and hyperbaric oxygen therapy

A normal healing wound is thought to progress through four stages; haemostasis, inflam-
mation, cell proliferation, and tissue remodelling (Thackham et al., 2008), although these
processes are interconnected and overlapping. During haemostasis, which lasts for sev-
eral hours, the blood flow is halted. The inflammation phase lasts for approximately one
week and sees the production of chemoattractants that stimulate the migration of fibrob-
lasts into the wound. Fibroblasts are the dominant cell type during the proliferative stage
of healing, which lasts several weeks. These cells produce collagen, the main compo-
nent of the extracellular matrix (ECM). The cocktail of chemoattractants also stimulate
the systematic rearrangement of endothelial cells (ECs) from neighbouring blood vessels
(Diegelmann and Evans, 2004). Capillary sprout extension is facilitated by EC prolifer-
ation and directed migration toward the chemical attractant. The fusion of two capillary
sprouts within a healing wound forms a loop through which blood can flow. Additional
new sprouts develop from this looped vessel, thus propagating angiogenesis. The final
stage of healing, tissue remodelling, can last for several months or even years.
The role of oxygen in wound healing is complex. Hypoxia (low oxygen) is required
to initiate angiogenesis (Gordillo and Sen, 2003). However, if the oxygen level is not
corrected and hypoxia persists, a chronic wound may form (Gordillo and Sen, 2003). It
has been said “although hypoxia can initiate neovascularisation by inducing angiogenic
factor expression, it cannot sustain it” (Gordillo et al., 2008). Hyperbaric oxygen therapy
(HBOT) is a treatment aimed at raising the oxygen levels within the wound bed which
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1869

involves exposing a patient to intermittent 100% oxygen at a pressure greater than 1 at-
mosphere (atm). Its use is supported in the treatment of conditions such as carbon monox-
ide poisoning, chronic wounds, and thermal burns (Gill and Bell, 2004). However, there
is much debate about the optimal HBOT protocol for treating chronic wounds (Tompach
et al., 1997; Niklas et al., 2004). Some treatments involve normobaric pressure while
some hyperbaric protocols raise the pressure up to 3 atmospheres with session durations
ranging from minutes to 3 hours (see, for example, Hammarlund and Sundberg, 1994;
Faglia et al., 1996; Abidia et al., 2003). Typically a patient receives one treatment per day
for about 6 weeks and the most common protocol involves exposure to oxygen at 2.4 atm
for 90 minutes a day. The need to identify the optimal and safe protocol(s) is empha-
sised by the fact that oxygen therapy has been associated with adverse side effects. For
instance, hyper-oxygenation is a therapy used in neonates even though it has been associ-
ated with childhood blindness (retinopathy of prematurity) and cancer (Silverman, 2004;
Sola et al., 2008). There are several adverse effects associated with this therapy including
oxygen toxicity, barotrauma, and decompression sickness (Williams, 1997).
HBOT increases the amount of oxygen delivered to the hypoxic wound site by in-
creasing the blood-oxygen level within the injury (Hammarlund and Sundberg, 1994;
Mathieu, 2002). Hehenberger et al. (1997) claim that the pro-healing effects of HBOT
stem from this increase in oxygen supply, while Chen et al. (2007) argue that the precise
cause is not well understood. Therapeutic effects, such as increased collagen synthesis
and improved bacterial killing have been reported during the periods of elevated oxy-
gen tension, while other effects, including down-regulation of inflammatory signals that
are overabundant in chronic wounds, may continue long after the treatment has ceased.
Thackham et al. (2008) provide a more detailed review of the wound healing process and
the use of HBOT to treat non-healing wounds.

1.2. Review of previous mathematical models

Pettet et al. (1996b) developed the first model of wound healing angiogenesis, using par-
tial differential equations (PDEs) to represent key components involved in the process.
Byrne and Chaplain (1996) developed and investigated a PDE model of tumour-induced
angiogenesis and derived necessary conditions in terms of the model parameters for suc-
cessful angiogenesis. Levine and coworkers considered continuous limits of reinforced
random walks in order to construct a model of the angiogenesis process (Levine et al.,
2000, 2001, 2002). Gaffney et al. (2002) applied travelling wave analysis of a simple
PDE model to determine a lower bound on the speed of healing in terms of the random
motility and budding of capillary tips. An alternative model of angiogenesis was devel-
oped by Sun et al. (2005) using a multiscale approach that coupled processes at the tissue
and cell scales.
In 1997, Olsen et al. (1997) developed a two-species model of wound healing an-
giogenesis in one and two dimensions. The two species were the density of ECs
forming newly developed capillaries in the wound and an insoluble ECM substrate.
Mechanochemical models of wound healing include the early work by Murray et al.
(1983, 1988), Tranquillo and Murray (1992) and the extensions by Olsen et al. (1995)
and Cook (1995). These models consider the connection of cells to the ECM and are thus
relevant for deeper, dermal wounds. They are typically applied to acute (normal) healing
wounds that heal primarily by contraction.
1870 Flegg et al.

More recently, Schugart et al. (2008) developed a model of acute wound healing an-
giogenesis that extends the work of Pettet et al. (1996b). The model accounts for in-
teractions between capillary tips, blood vessels, oxygen, inflammatory cells, chemoat-
tractant, fibroblasts, and ECM during the acute healing process. The impact of oxygen
therapy on the total wound EC densities is assessed. Under normoxia and hyperoxia
(as during oxygen therapy), it is assumed that proliferation of fibroblasts and endothe-
lial cell sprouts increases linearly with the local oxygen concentration. Chemoattractant
production is also assumed to be upregulated under hypoxia and hyperoxia and down-
regulated under normoxia. An increase in chemoattractant production during hyperoxia
is supported by some researchers (Polverini, 2002; Al-Waili and Butler, 2006) although
there are those that claim otherwise (Raa et al., 2007). Schugart et al. (2008) predict
a positive effect on wound healing under oxygen therapy and a negative effect under
excessive treatment. They also predict that healing will continue under excessive oxy-
gen exposures, albeit more slowly than under normal conditions. These results are in
disagreement with the work of Knighton et al. (1981) who state that excessive expo-
sure would likely halt the healing process altogether. Schugart et al. (2008) conclude
that extreme hypoxia is detrimental to wound healing while hyperoxia promotes heal-
ing.
Recently, Xue et al. (2009) extended the work by Schugart et al. (2008). Their model
consists of eight interacting chemical and cell species, namely, ECM, oxygen, platelet
derived growth factor (PDGF), vascular endothelial growth factor (VEGF), macrophages,
fibroblasts, capillary tips, and capillary sprouts. They view the ECM as a viscoelastic
material and include two chemoattractants to account for the fact that PDGF acts as an
attractant for macrophages and fibroblasts while VEGF acts to initiate angiogenesis. It is
the first mechanochemical model of wound healing angiogenesis.
The model presented in this paper is designed to investigate the mechanisms by which
HBOT may improve wound healing. Since its benefits are thought to stem from the physi-
ological and pharmacological effects of increasing oxygen provided to the wound (Gajen-
drareddy et al., 2005), we focus on the impact of the increased concentration of oxygen
that is supplied during HBOT. Our model is distinct from Schugart et al. (2008) for a va-
riety of reasons. Firstly, we are interested in the application of HBOT to chronic, diabetic
wounds. Secondly, Schugart et al. (2008) do not distinguish between different physically
realistic protocols, whereas we attempt to do so. Finally, in our model the equilibrium
blood vessel density depends on the ECM density (since the ECM contains cells that con-
sume nutrient; at steady state the blood vessel density will be such that the metabolic
needs of the tissue are met). In particular, feedback between the cellular components of
tissue and its vascular density regulate the vascular remodelling process. The work of Xue
et al. (2009) does not investigate the treatment of non-healing wounds, which is the focus
of our paper.
The remainder of the paper is organised as follows. In Section 2, the mathematical
model including the governing equations, boundary conditions, and initial conditions are
introduced. Numerical simulations are presented in Section 3, followed by a detailed dis-
cussion of their significance in Section 4, where we also discuss future directions for this
research.
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1871

2. Model development

We develop a system of PDEs that describe the interactions in space and time of biological
factors that regulate the healing process. We present the governing equations in a general
3-dimensional framework, where x represents position and t denotes time. Six species
are considered; c, oxygen concentration; a, chemoattractant concentration; n, capillary
tip density; b, blood vessel density; f , fibroblast density and e, ECM density. We present
the governing equations first and then explain and justify the terms that appear in each
equation.
Oxygen, c(x, t):

∂c (λ1 + λ2 e)c  
= Dc ∇ 2 c − − λ4 bc + λ5 b 1 + αg(t) , (1)
∂t diffusion λ3 + c removal by supply from vessels and
consumption by tissue vessels HBOT-induced supply
and decay

where g(t) = 1 while the patient is undergoing HBOT, otherwise g(t) = 0.

Chemoattractant, a(x, t):
∂a λ8 G(c)
= Da ∇ 2 a − λ6 ab − λ7 a + , (2)
∂t diffusion removal by natural e0 + e
vessels decay production

where G(c) = 1 if c ∈ [cdwn , cup ], otherwise G(c) = 0.

Capillary tips, n(x, t):
 
∂n −χn en
=∇· ∇a + a(λ9 b + λ10 n)
∂t (e02 + e2 )(γ + a)2 sprouting
chemotaxis

− n(λ11 n + λ12 b) . (3)

anastomoses

Blood vessels, b(x, t):

 
∂b −χn en  
= ∇a · v̂1 + λ13 b [λ14 e + λ15 f ] − b , (4)
∂t (e0 + e )(γ + a)
2 2 2
remodelling
snail trail production
from capillary tips

where v̂1 is a unit vector pointing in the direction of the leading capillary tip (Pettet et al.,
1996b). In the case of a one-dimensional wound that travels from right to left, v̂1 = −i.
Fibroblasts, f (x, t):
 
∂f −χf f λ16 f c λ17 f 2
=∇· ∇a + − . (5)
∂t (γ + a)2 c0 + c (c0 + c)(e0 + e)
chemotaxis proliferation death

ECM, e(x, t):

∂e
= λ18 f c(λ19 c − e) . (6)
∂t deposition
1872 Flegg et al.

In Eqs. (1)–(6), λi (i = 1, . . . , 19), e0 , γ , c0 , χn , χf , α, Dc and Da are positive constants.

We will now discuss each of Eqs. (1)–(6).

Consider Eq. (1). When vascularised tissue is wounded, blood vessels that previously
supplied the tissue with oxygen are severed. At this time, an oxygen gradient is estab-
lished whereby the oxygen concentration in the neighbouring healed tissue is at “normal”
levels (that is, the level found in healed tissue) while the oxygen concentration within the
wounded space is low (David et al., 2001). The gradient leads to diffusion of oxygen from
the vascularised region into the wound space. We therefore incorporate diffusion, with
diffusivity Dc , into our governing equation for the evolution of the oxygen concentration.
Following Bauer et al. (2005b), new blood vessels that have been laid down behind the
migrating capillary tips are assumed to supply the developing wound space with oxygen
at a constant rate λ5 . Sheffield (1985) found that the oxygen tension rises quickly when
HBOT commences and falls rapidly when the treatment stops. HBOT is therefore assumed
to increase the oxygen supply rate from existing vessels to λ5 (1 + α) during a HBOT
session. The parameter α is a measure of the augmentation of the blood oxygen level and
the “strength” of the HBOT session: It is discussed later.
Since vessels are known to supply and remove oxygen from the tissue (Dewhirst,
1994), we include in our model a term that represents the removal of oxygen by the
vasculature, at rate λ4 . Oxygen is assumed to decay at rate λ1 /(λ3 + c) and be consumed
by the wound tissue (Gottrup, 2004) at rate λ2 e/(λ3 + c). These rates are both limited
as oxygen levels rise indefinitely. The consumption rate is proportional to ECM density
so that as ECM levels rise (that is, there is more granulation tissue in the wound), more
oxygen is consumed.

Consider Eq. (2). It is important to note that the oxygen gradient has been shown to
recruit macrophages into the wound space (David et al., 2001). Macrophages play an im-
portant role in wound healing, clearing debris, and releasing chemical stimulants such
as VEGF (Cliff, 1963; Harmey et al., 1998; Tandara and Mustoe, 2004). We avoid ex-
plicitly including a macrophage population in our model by noting that the oxygen level
and macrophage density are inter-related. In particular, we assume that the production of
chemoattractant by macrophages occurs for a finite range of the local oxygen concentra-
tion, c ∈ [cdwn , cup ] and at a rate λ8 /(e0 + e), which depends on the local ECM density.
The rate of production of chemoattractant is decreased in regions of increased ECM den-
sity since these are the regions where the wound healing unit has already passed through
and the inflammatory process has subsided. Such feedback ensures that the angiogenesis
response weakens as new collagen is deposited (Acker et al., 2001). It has been confirmed
that a minimum level of oxygenation is needed to facilitate wound healing (Gordillo et al.,
2008), which further supports our use of a lower bound on the oxygen concentration for
chemoattractant production. Once produced, the chemoattractant diffuses from regions
of high concentration to low concentration and in our model, with diffusivity Da . The
chemoattractant is removed from the wound space via natural decay at rate λ7 and via the
vasculature at rate λ6 (Serini et al., 2003).

Consider Eq. (3). The chemoattractant stimulates the proliferation of ECs in the vessels
of nearby healthy tissue, their directed movement away from the parent vessel (Bauer et
al., 2005b), and their alignment into capillary tips (Diegelmann and Evans, 2004). We
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1873

model this process by following Keller and Segel (1971), and assuming that the chemo-
tactic sensitivity depends on the chemoattractant concentration, a. A variety of choices for
χn (a) have appeared in the literature. For instance, Pettet et al. (1996b) used the “receptor-
kinetic” form:
χn
χn (a) = ,
(γ + a)2

while the simplest case of χn (a) = χn has been extensively used (Pettet et al., 1996a;
Byrne and Chaplain, 1996; Chaplain and Orme, 1998; Byrne et al., 2000). We choose the
receptor-kinetic form which produces a stronger chemotactic response at low concentra-
tions of chemical (when the number of unoccupied receptors for the chemoattractant is
high) and a lower response when the concentration is high (and there are few unoccupied
receptors). However, sufficient ECM is necessary to support EC ingrowth and to serve as
a scaffold for EC migration (Olaso et al., 2002; Carmeliet, 2004). We therefore model the
chemotactic movement of capillary tips into the wound space as depending on both the
chemoattractant concentration, a and the local ECM density, e. We take

χn e
χn (a, e) = ,
(e02 + e2 )(γ + a)2

where the (γ + a)−2 component represents the receptor-kinetic law. The factor e
(e02 +e2 )
ensures that, while ECM is required for EC migration into the wound space, excessive
ECM deposition (as in the case of keloid and hypertrophic scars (Alessio et al., 1998;
Ichioka et al., 2008)) will lead to limited capillary migration due to space restrictions.
The processes of capillary sprouting and anastomoses are important in wound healing
(Cliff, 1963) and therefore both are included in the equation for the capillary tips. Accord-
ing to Adams and Alitalo (2007), in order for new vascular connections to form, tip cells
need to encounter their targets, the tips of other sprouts or existing capillaries. We model
this by including terms λ11 n2 for tip-tip connections and λ12 nb for tip-sprout connections.
We include sprouting from both blood vessels λ9 ab and capillary tips λ10 an, but each
process is assumed to be proportional to the chemoattractant concentration (Polverini,
2002; Bauer et al., 2005a; Adams and Alitalo, 2007).

Consider Eq. (4). The “snail-trail” model has been widely used in situations where one
species or cell directly follows another which is undergoing chemotaxis (Edelstein, 1982;
Edelstein et al., 1983; Edelstein and Segel, 1983; Balding and McElwain, 1985). In par-
ticular, it has been applied extensively to models of wound healing and tumour-induced
angiogenesis (see, for example, Pettet et al., 1996a, 1996b; Byrne and Chaplain, 1996;
Byrne et al., 2000; Panovska et al., 2008). The snail-trail approach asserts that if the capil-
lary tips, n, move with a velocity v, then the rate of increase (that is, production/extension)
of vessels is given by (nv) · v̂, where v̂ is a unit vector in the direction of v. Given that the
flux of capillary tips is

−χn e
J= ∇a,
(e02 + e2 )(γ + a)2
1874 Flegg et al.

then the production term for blood vessels takes the form
 
−χn en
∇a · v̂1 ,
(e02 + e2 )(γ + a)2

where v̂1 is a unit vector pointing in the direction of the leading capillary tip.
The conversion of capillary tips into blood vessels can lead to overproduction of ves-
sels, which will eventually be remodelled and regress (Polverini, 1995). We model re-
gression as logistic growth at rate λ13 with a carrying capacity that depends on both the
underlying ECM concentration (λ14 e) and the fibroblast density (λ15 f ). This ensures that,
at equilibrium, the vessel density will meet the metabolic demands of the fibroblasts and
ECM that constitute the tissue (Gordillo et al., 2008).

Consider Eq. (5). The presence of a chemoattractant gradient in the wound site stim-
ulates fibroblast migration (Oberringer et al., 2007). Hence, we assume that fibroblasts
undergo chemotaxis up the chemoattractant gradient with chemotactic sensitivity

χf
χf (a) = ,
(γ + a)2

where the (γ + a)2 component represents the receptor-kinetic as described above.

Fibroblast proliferation has been observed to increase with increasing oxygen levels
(Kessler et al., 2003; Lin et al., 2008) and so we employ a proliferation rate, cλ016+cc , which is
an increasing, saturating function of oxygen. Fibroblasts die (Clark, 1996) and we assume
that their death rate is given by,

λ17 f
,
(c0 + c)(e0 + e)

which increases under hypoxia. We also assume that the local ECM density influences the
rate of fibroblast death, so that in regions where ECM levels are low (that is, in front of
the wound healing unit) the rate of fibroblast death will be maximal.

Consider Eq. (6). It is known that fibroblasts produce collagen which is a major com-
ponent of the ECM (Trabold et al., 2003). The rate at which fibroblasts synthesis collagen
is known to be an increasing function of the oxygen concentration (Fries et al., 2005).
To represent these inter-relationships, we use a logistic growth term in the equation gov-
erning the evolution of the ECM density, where the rate of deposition is proportional to
both the fibroblast density and oxygen concentration. The use of a logistic growth term
to model ECM deposition by fibroblasts is common in the literature (Pettet et al., 1996b;
Olsen et al., 1997, 1998). Gordillo et al. (2008) state “a threshold level of oxygenation is
required to support the metabolic demands of tissue remodelling.” We therefore take the
carrying-capacity of the ECM to depend on oxygen concentration. That is, under steady
state conditions, the oxygen concentration must be sufficient to meet the metabolic needs
of the tissue.
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1875

Table 1 HBOT protocols to be tested using the model. Clinical data is from Piantadosi (1999)

Pressure Oxygen PAO2 PAO2 α

(atm) level calculated clinical

1 21% 100 89 0
1 100% 673 507 5.73
2 100% 1433 N.A. 13.33
2.4 100% 1737 N.A. 16.37
3 100% 2193 1721 20.93

2.1. Boundary and initial conditions

While Eqs. (1)–(6) are stated in a general three-dimensional form, in what follows we
restrict attention to a one-dimensional Cartesian geometry in which x denotes distance
from the wound centre (at x = 0). We assume that the wound is symmetric about x =
0 and initially of length 2L so that (at t = 0) the wound margins lie at x = ±L. We
consider the space between L and 2L (and also between −L and −2L due to symmetry)
to be healed tissue that surrounds the wound. The dimensionless model equations and
associated boundary and initial conditions are presented in Appendix A. Here, we assume
that initially the region 0 ≤ x ≤ L represents the wounded tissue and that the region L <
x ≤ 2L represents healed tissue. Additionally, we do not explicitly track the interface (or
moving boundary) that separates the wounded and healed tissues.

2.2. Determination of the parameter α: the effective increase in oxygen at the local level
due to systemically applied HBOT

In the absence of clinical data on the increase in oxygen tensions under HBOT, we use a
modelling approach, supported by clinical observations, to estimate the parameter α. The
(calculated mean) alveolar oxygen tension (PAO2 )1 is given by Piantadosi (1999) as
 
1 − FIO2
PAO2 = FIO2 (Pb − PH2 O ) − PACO2 FIO2 + , (7)
R

where FIO2 is the inspired oxygen level as a percentage, Pb is the barometric pressure
(1 atm = 760 mmHg), PH2 O is the water vapour pressure (47 mmHg at 37°C), PACO2 is
the mean alveolar carbon dioxide tension (40 mmHg) and R is the respiratory quotient
(0.8 when breathing air and 1.0 when breathing 100% oxygen). This formula predicts
that if a patient is breathing 21% oxygen at 1 atm then the alveolar oxygen tension is
PAO2 = 0.21(760 − 47) − 40[0.21 + (1 − 0.21)/0.8] ≈ 100 mmHg. In previous work, we
combined a mathematical model of wound healing with Eq. (7) to estimate the parameter
α associated with typical HBOT protocols (Flegg et al., 2009). Table 1 summarises the
estimates of α that we use in our simulations.

1 Note that this is not an empirical formula, rather it is based on the use of an ideal gas law.
1876 Flegg et al.

2.3. The impact of diabetes on the wound healing environment

The ways in which diabetes affect wounds are complex and not well understood. We
outline below some of the clinical and experimentally observed features that we use our
model to investigate.
1. Decreased or insufficient oxygen supply. Oyibo et al. (2001) state that the outcome of
diabetic foot ulcers is influenced by (at least) blood supply and the presence of infec-
tion. An adequate oxygen supply is essential for healing and an inadequate oxygen
supply is often reported as a cause of wound chronicity (Hunt et al., 1967, 1969; Sen
et al., 2002; Said et al., 2005; Hopf et al., 2006; Lin et al., 2008). We model decreased
oxygen supply by reducing the parameter λ5 in Eq. (1).
2. Increased bacterial load and/or infection. The ability to kill bacteria is reduced in
chronic wounds (Chang et al., 1983; Allen et al., 1997): this can overwhelm the wound
space and cause chronic hypoxia. Chen et al. (1999) state that bacterial infection is one
of the major etiologies of non-healing wounds. When inflammatory cells kill bacteria
present in the wound, large amounts of oxygen are consumed. In fact, Cianci (2004)
state that phagocytosis can stimulate up to a five fold increase in oxygen consumption.
We account for a higher bacterial load by increasing λ1 , the rate at which oxygen is
consumed in Eq. (1).
We note that as a wound heals and bacteria is cleared, the rate of oxygen consump-
tion should decline to normal levels. To model this, we could introduce a new variable
to represent the bacterial load. The governing equation for oxygen (see Eq. (1)) would
then be modified to include an extra component to model oxygen consumption by bac-
teria while bacteria load would decrease at a rate which may depend on the rate of
oxygen consumption.
3. Insufficient blood flow. According to Hammarlund and Sundberg (1994), impaired
wound healing is often associated with a compromised or reduced blood supply and
we model this effect by reducing either one or both of the parameters λ14 and λ15 in
Eq. (4).
4. Disruption of normal ECM deposition. Insufficient deposition of ECM is thought to
cause stalled healing while excessive deposition can cause fibrosis (Diegelmann and
Evans, 2004). We account for a lack of ECM by reducing λ19 (to reduce the steady
state levels) and λ18 (to reduce the rate at which ECM is laid down) in Eq. (6). Fibrosis
is simulated by increasing these parameters.
5. Failed vascular networking. It is known that chronic hypoxia inhibits collagen synthe-
sis and angiogenesis (Hollander et al., 2000; Hopf et al., 2005; Slovis, 2006). Dor et
al. (2003) state that insufficient VEGF can result in regression of newly acquired ves-
sels. In Eq. (3), we model the inhibition of angiogenesis by decreasing parameters λ9
and/or λ10 and the regression of new vessels (that is capillary tips) by increasing λ11
and/or λ12 .
6. Decreased fibroblast proliferation. While it has been reported that the rate at which
diabetic fibroblasts proliferate is not significantly lower than that of normal fibrob-
lasts (Lerman et al., 2003), some authors claim that delayed and decreased fibrob-
last proliferation is a characteristic feature of nonhealing wounds (Moore et al., 1997;
Enoch and Price, 2004; Lin et al., 2008). In Eq. (5), we model decreased fibroblast
proliferation by reducing the parameter λ16 .
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1877

Table 2 Parameter values for the normal wound simulation in Fig. 1(a)

k1 = 50 k2 = 50 k3 = 5 k4 = 25 k5 = 50 k6 = 10
k7 = 10 k8 = 50 k9 = 10 k10 = 10 k11 = 25 k12 = 25
k13 = 25 k14 = 0.25 k15 = 0.1 k16 = 3 k17 = 10 k18 = 30
k19 = 1 κa = 0.25 κf = 1 κn = 1 cH = 0.7 cL = 0.2

7. Reduced production of growth factors. Lerman et al. (2003) state that diabetic fibrob-
lasts produce significantly less VEGF than normal fibroblasts. Furthermore, Mace et
al. (2007) note that “impaired wound healing in diabetic patients is associated with
deficiencies in the production of factors involved in cell proliferation and migration
such as VEGF” and that this down-regulation is associated with hypoxia. We model
this effect by reducing the parameter λ8 in Eq. (2).
8. Decreased chemotactic responsiveness. It has been observed that fibroblast migration
is the rate-limiting step in granulation tissue formation (Clark et al., 2007). Accord-
ing to Lerman et al. (2003), diabetic fibroblasts are less motile than their wild-type
counterparts. Macrophages and neutrophils have been observed to exhibit similar be-
haviour (Eginton et al., 2003; Simons, 2005). Further, Singer and Clark (1999) state
that diabetic ulcers have impaired granulocytic function and chemotaxis. We model the
reduced chemotactic responsiveness of fibroblasts and ECs by reducing the parameters
χf and χn in Eqs. (3), (4), and (5).
For each of the potential causes of stalled wound healing mentioned above, we use our
model to assess whether HBOT will assist healing in a significant way. In doing so, we
aim to identify those patients whose wounds will respond positively to HBOT.

3. Numerical results

After appropriate rescaling (see Appendix A), Eqs. (1)–(6) were solved numerically using
a flux limiting finite volume approach (see Thackham et al., 2009 for more details). Ta-
ble 2 shows the dimensionless parameter values used to simulate normal healing. They are
based on values used by Pettet et al. (1996b). We focus on generating qualitative simula-
tions of wound healing and qualitative insight into the likely response of different types of
wounds to treatment with HBOT. Our justification for focusing on qualitative rather than
quantitative aspects centres around the large number of parameters that appear in models
of wound healing such as ours and the absence of suitable data with which to estimate
many of them.
In Fig. 1(a), we present numerical results which show how the model variables are
distributed within a normal wound after 3 weeks of healing. The fibroblasts have deposited
ECM which has allowed capillary tips to lay down blood vessels throughout most of
the wound. The chemoattractant has initiated angiogenesis and restored oxygen to near
normal levels throughout most of the wound.
As mentioned above, we have not attempted to estimate the parameters in our model
from clinical and/or experimental data. We have performed a parameter sensitivity analy-
sis (see Table 3). While many of the changes are as expected, some are not. For example,
decreasing or increasing either the rate of oxygen consumption (k1 ) or the rate of oxygen
1878 Flegg et al.

Fig. 1 (a) Numerical results from a typical simulation showing how the six components of our model
are distributed within a normal wound after 3 weeks of healing. (b) Numerical results showing how the
normal wound in (a) responds to 3 weeks of HBOT (2.4 atm, 90 minutes per day). Key: oxygen (gray dot),
chemoattractant (gray dash), capillary tips (gray), blood vessels (black dot), fibroblasts (black dash), and
ECM (black).

supply (k5 ) decreases the rate at which healing occurs. Changing the rate of oxygen sup-
ply or oxygen consumption from their baseline values by a factor of 10 causes delayed,
stalled or no healing at all since the oxygen levels are not conducive to chemoattractant
production (see the G(c) function in Eq. 3).
Many hyperbaric centres advocate the use of HBOT to treat ‘normal’ wounds and to
accelerate healing in sports injury (James et al., 1993). This approach is highly contro-
versial (Babul and Rhodes, 2000) and there is little evidence to support the use of HBOT
for treating normal wounds. Here, we use our model to investigate the impact of apply-
ing HBOT to a normal healing wound; the protocol involving exposure at 2.4 atm for 90
minutes each day. While Fig. 1(b) shows that HBOT can marginally assist the closure of
normal healing wounds (a 10% increase in the speed of healing, at most), in practice its
high cost would have to be taken into consideration when assessing whether to use it to
treat such wounds. Alternatively, mechanisms not included in our model, such as nitric-
oxide production, are up-regulated by HBOT (Boykin and Baylis, 2007). Nitric-oxide can
promote wound angiogenesis by inducing expression of VEGF (Sen et al., 2002).
In Section 2.3, we introduced a number of possible causes for stalled healing. We
focus our investigation on the effect that HBOT has on a wound that has stalled due
to excessive bacteria. We model this effect by increasing the rate of oxygen consump-
tion from k1 = 50 to 200 (Cianci, 2004). Figure 2(a) shows the spatial distribution of
the six components of our model within such a stalled, chronic wound 3 weeks after the
wound was created. The wound fails to heal and hypoxia effects the chemoattractant pro-
file (since the oxygen levels are too low to allow chemoattractant production) which, in
turn, disrupts the healing progress. Figure 2(b) shows how such a wound responds to 3
weeks of treatment with HBOT (100% oxygen and 2.4 atm for 90 minutes per day). The
additional supply of oxygen delivered to the wound by HBOT has initiated chemoattrac-
tant production (see the G(c) function in Eq. 3), which ultimately allows the wound to
heal.
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1879

Table 3 Summary of results from parameter sensitivity analysis, for the normal wound parameter values
shown in Table 2. All simulations were run for 6 weeks. Wound healing times marked with a ‡ were
extrapolated from the healing speed over the first 6 weeks. All parameter values for which the results of
the parameter sensitivity analysis are counter intuitive have a † placed beside them

Parameter and Parameter Wound healed Parameter Wound healed

baseline value change by change by

Normal parameters – 23 days

†k1 = 50 k1 = 500 No healing occurs k1 = 5 ‡60 days
k2 = 50 k2 = 500 No healing occurs k2 = 5 24 days
†k3 = 5 k3 = 50 Healing stalls k3 = 0.5 No healing occurs
k4 = 25 k4 = 250 No healing occurs k4 = 2.5 18 days
†k5 = 100 k5 = 1000 Healing stalls k5 = 10 No healing occurs
k6 = 10 k6 = 100 26 days k6 = 1 23 days
†k7 = 10 k7 = 100 ‡62 days k7 = 1 28 days
†k8 = 50 k8 = 500 31 days k8 = 5 ‡54 days
k9 = 10 k9 = 100 18 days k9 = 1 26 days
k10 = 10 k10 = 100 11 days k10 = 1 ‡44 days
k11 = 25 k11 = 250 ‡43 days k11 = 2.5 22 days
k12 = 25 k12 = 250 29 days k12 = 2.5 21 days
k13 = 25 k13 = 250 19 days k13 = 2.5 22 days
k14 = 0.25 k14 = 2.5 11 days k14 = 0.025 Wound area increases
k15 = 0.1 k15 = 1 14 days k15 = 0.01 26 days
k16 = 3 k16 = 30 16 days k16 = 0.3 30 days
k17 = 10 k17 = 100 ‡45 days k17 = 1 20 days
k18 = 30 k18 = 300 21 days k18 = 3 ‡44 days
k19 = 1 k19 = 10 19 days k19 = 0.1 Wound area increases
κa = 0.25 κa = 0.5 24 days κa = 0.125 23 days
κn = 1 κn = 2 15 days κn = 0.5 37 days
κf = 1 κf = 2 27 days κf = 0.5 23 days
†cH = 0.7 cH = 1.4 29 days cH = 0.35 ‡60 days
cL = 0.2 cL = 0.4 33 days cL = 0.1 22 days

The results presented in Fig. 3 show how normal and chronic wounds respond to treat-
ment, over a period of 5 weeks, with HBOT administered at 100% oxygen and 2.4 atm
for 90 minutes per day. Wound area is measured by finding the location where the blood
vessel density reaches a value of 0.05 and then calculating area as the square of the re-
maining wound tissue. Several predictions can be made from the results shown in Fig. 3.
Firstly, supernormal healing is not stimulated by the physiological HBOT protocol con-
sidered here. That is, the normal wound area (shown in solid gray) is always less than
the chronic wound under treatment (shown in solid black). This conclusion is consis-
tent with the literature (Williams, 1997). Secondly, treatment of non-healing wounds with
HBOT must continue until healing is complete otherwise healing will stall; this agrees
with the observations of Niklas et al. (2004). Healing stalls because when the treatment
finishes prematurely, the oxygen concentration in the wound margin (L ≤ x ≤ 2L) allows
1880 Flegg et al.

Fig. 2 (a) Numerical results showing how the six components of our model are distributed within a
chronic wound after 3 weeks of healing. In this case, wound healing stalls due to excessive bacteria which
causes a four-fold increase in the oxygen consumption rate (that is, k1 increases from k1 = 50 in Fig. 1 to
k1 = 200). (b) Numerical results showing how the chronic wound considered above responds after 3 weeks
of HBOT (2.4 atm, 90 minutes per day). Key: oxygen (gray dot), chemoattractant (gray dash), capillary
tips (gray), blood vessels (black dot), fibroblasts (black dash), and ECM (black).

Fig. 3 Results from a series of simulations showing how the healing of normal and chronic wounds is
affected by treatment with HBOT. In each case, we plot the time evolution of the relative wound area
(wound area/initial wound area) over a period of 5 weeks. Key: chronic wound without HBOT (black
dash), chronic wound with HBOT (black solid), chronic wound with HBOT stopped after 2 weeks of
treatment (black dot), normal wound without HBOT (gray solid), and normal wound with HBOT (gray
dash). Parameter values for the normal and chronic wounds are as in Figs. 1 and 2, respectively.

chemoattractant production behind the healing unit, which stalls the healing process. If
we apply HBOT for four weeks, then complete healing is achieved.
Due to the relatively expensive nature of HBOT and the lack of evidence supporting
its cost-effectiveness in treating chronic wounds (Leach et al., 1998; Thom, 2009), cost
savings may be made if patients receive fewer, longer sessions. Furthermore, patients
often complain that attending treatment sessions every day is quite inconvenient (Woo
et al., 1997). Reducing the frequency of sessions would decrease the inconvenience and
travel time for patients. On the other hand, more frequent sessions of shorter duration
may significantly improve healing. We can use our mathematical model to investigate and
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1881

Fig. 4 Results from a series of simulations showing how the healing of a chronic wound is affected by
different HBOT protocols. In each case, we plot the time evolution of the relative wound area over a period
of 3 weeks. Key: chronic wound without HBOT (gray dash), chronic wound with 45-minute sessions twice
per day (black dot), chronic wound with 90-minute sessions once per day (black dash), chronic wound with
3-hour sessions every second day (gray solid) and chronic wound with 6-hour sessions one per week (black
solid). Parameter values for the chronic wound are as in Fig. 2.

compare the efficacy of different HBOT protocols. In Fig. 4, we use our mathematical
model to compare four different HBOT protocols by monitoring the wound area over a
three week period. For each protocol, the total weekly exposure to HBOT is comparable.
Based on these results, we can draw a number of important and clinically relevant con-
clusions. Firstly, treatment with two 45-minute sessions per day is only marginally better
than treatment with a single, 90-minute session. In this case, the inconvenience to the pa-
tient of having to return to a clinic twice in one day would probably outweigh the slight
improvement in the healing rate. Secondly, a six-hour session once per week is completely
ineffective in assisting wound closure and should not be considered as an alternative to
the standard protocol. This is consistent with reports that sessions of more than 3 hours
can be detrimental to healing (Dimitrijevich et al., 1999). Finally, a protocol involving 3-
hour sessions every second day may be worth considering for patients who need to travel
long distances in order to receive treatment since the final wound area after 3 weeks of
treatment is only marginally greater than either of the more frequent treatments.
Figure 5 shows how the concentration of oxygen at the wound edge changes with
time during the first week for three HBOT protocols shown in Fig. 4. Our numerical
simulations predict that the oxygen concentration rises rapidly when a treatment session
starts and falls rapidly after a session finishes; these results are consistent with the clinical
observations made by Sheffield (1985). From Figs. 4 and 5, we can conclude that the
intermittent oxygen exposure associated with HBOT is crucial for stimulating healing in
a chronic wound.
We now compare HBOT protocols which differ in duration and the applied pressure.
Figure 6 shows how the area of a chronic wound after 3 weeks of HBOT depends on
the protocol used. HBOT is administered daily, for a given time period (0, 30, 60, 90,
120, 150, or 180 minutes) over a 3-week period, for each of the five protocols listed in
Table 1. Several predictions can be made from our results. Firstly, while oxygen therapy
(that is treatments using 100% oxygen) at any pressure and of any duration assists the
1882 Flegg et al.

Fig. 5 Results from a series of simulations showing how the concentration of oxygen at the edge of the
chronic wound is affected by different HBOT protocols. In each case, we plot the time evolution of the
oxygen concentration at the wound edge over a period of 7 days. Key: chronic wound with 45-minute
sessions twice per day (gray solid), chronic wound with 90-minute sessions once per day (black dash),
and chronic wound with 3-hour sessions every second day (black solid). Parameter values for the chronic
wound are as in Fig. 2.

Fig. 6 Results from a series of simulations showing how the relative wound area is affected by different
HBOT protocols. In each case, we display the relative wound area after a period of 3 weeks, for six
different durations of exposure (0, 30, 60, 90, 120, 150, and 180 minutes). Key: chronic wound without
HBOT (solid black), chronic wound treated with 100% oxygen at 1 atm (vertical lines), chronic wound
treated with 100% oxygen at 2 atm day (solid white), chronic wound treated with 100% oxygen at 2.4 atm
(dotted lines), and chronic wound treated with 100% oxygen at 3 atm (horizontal lines). Parameter values
for the chronic wound are as in Fig. 2.

healing of chronic wounds, higher pressures, and longer sessions have a greater impact
on increasing the healing rate. Thus our simulations suggest that normobaric oxygen is
not a viable substitute for HBOT, a conclusion which is borne out by Lin et al. (2008) who
observed that a patient with a compromised wound who is breathing normobaric oxygen
(100% oxygen at 1 atm) may not experience a sufficient pressure gradient to drive oxygen
to reach the cells.
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1883

Fig. 7 Results from a series of simulations that compare the response of chronic wounds of different
etiologies to treatment with a standard HBOT protocol (100% oxygen and 2.4 atm for 90 minutes per day)
and the control (untreated healing). Key: unhealed chronic wound (gray) and HBOT-treated chronic wound
(black). Please note that for the two bar graphs with a ∗ placed at the top of them, an increase of the wound
was observed.

We also note from Fig. 6 that there are several protocols (of those considered in this
paper) that can stimulate healing of a chronic wound at a rate that is as fast, or faster,
than the healing rate of a normal wound. From Fig. 3, we can see that for the normal
wound, the relative wound area after 3 weeks is close to zero. The protocols that result
in healing rates which are as fast as, or faster than, normal healing are 100% oxygen at
3 atm for 1.5 or 2 or 2.5 or 3 hours a day and 100% oxygen at 2.4 atm for 2 or 2.5 or
3 hours a day and 100% oxygen at 2 atm for 2.5 or 3 hours per day. However, none of
these protocols are what would be considered “clinically reasonable” and we therefore
conclude that reasonable HBOT protocols cannot stimulate super-normal healing.
We now investigate the effect of HBOT on chronic wounds of different etiologies, as
described in Section 2.3. Figure 7 shows the wound area after 3 weeks for chronic wounds
of different origins under no treatment and the treated chronic wounds under a 2.4 atm and
90-minutes protocol per day. Identifying patients who are likely to benefit from HBOT be-
fore treatment is administered is important (Fife et al., 2002; Cianci, 2004). Here, we seek
to determine which types of chronic wounds will respond to HBOT. A summary of the pa-
rameter changes made to simulate each type of wound chronicity is given in Table 4. We
note that not all wounds benefit from HBOT. For example, the healing of chronic wounds
characterised by reduced chemotactic responsiveness, excessive ECM deposition, and/or
low chemoattractant production do not improve following treatment with HBOT. Since
high levels of ECM are often associated with hypertrophic scars (Alessio et al., 1998;
Ichioka et al., 2008), applying HBOT to such wounds may retard the hyper-proliferative
activity and allow a more mature collagen matrix to be established. However, keloid and
hypertrophic scars may be hypoxic due to reduced diffusion of oxygen from the microves-
sels to the tissues (Ichioka et al., 2008); these effects need to be incorporated into our
model before more detailed predictions can be made.
1884 Flegg et al.

Table 4 Changes in parameters values to simulate different wound chronicities

Cause of wound chronicity Parameter changes

Low chemotactic responsiveness κn = κf = 0.1

Low oxygen supply k5 = 50
Low blood vessels k14 = 0.025, k15 = 0.01
Low fibroblast proliferation k16 = 0.3
High anastomoses k11 = k12 = 250
Low budding/sprouting k9 = k10 = 1
High ECM deposition k18 = 100, k19 = 5
Low ECM deposition k18 = 5, k19 = 0.25
Low chemoattractant production k8 = 5
High oxygen consumption k1 = 200

4. Discussion

We have developed a mathematical model that simulates the healing of normal and
chronic wounds. Our model extends the previous work of Pettet et al. (1996b) and
Schugart et al. (2008) and is the first to simulate the application of hyperbaric oxygen
therapy to diabetic wounds.
Our simulations of chronic wounds and their treatment with HBOT have revealed that
the pro-healing effects of HBOT are not simply due to an overall increase in chemoat-
tractant production but also an altered spatial distribution of the chemoattractant con-
centration. The application of HBOT allows oxygen levels to return to normal and sub-
sequently the inflammatory response (as measured by chemoattractant concentration)
in the wounded region to subside. The concentration of the chemoattractant in this re-
gion then falls and a gradient conducive to chemotaxis is established. Fibroblast migra-
tion and ultimately angiogenesis can then occur in a region that would otherwise have
remained avascular (compare Figs. 2(a) and (b)). The concept that chronic inflamma-
tion is caused by an unfavourable spatial distribution of chemoattractant, rather than a
lack of chemoattractant, per se, is consistent with recent literature (Agren et al., 2000;
Chen and Rogers, 2007) and with the following statement by Gordillo et al. (2008): “Heal-
ing responses induced by HBOT occurred by means that were independent of VEGF ex-
pression.” VEGF is a growth factor that stimulates angiogenesis (Acker et al., 2001). Our
mathematical simulations support the hypothesis that stimulating chemoattractant pro-
duction may not be the cause of the benefit observed under HBOT; rather the change in
spatial distribution of the chemoattractant is the origin of the pro-healing effect. There is
complimentary evidence that HBOT influences nitric-oxide regulation and recruitment of
bone marrow derived endothelial progenitor cells to the wound site and may, in this way,
contribute to healing (Gallagher et al., 2006; Boykin and Baylis, 2007). We postpone
incorporation of these factors into our model for future work.
One of the main advantages of our modelling approach is that we can evaluate and
compare other possible treatments for chronic wounds including, for example, topical
oxygen therapy (Sen et al., 2002) and combination therapies involving HBOT and the
administration of either chemoattractants or vasodilators. In fact, in practice patients re-
ceiving HBOT will also receive other treatments such as wound dressings and therapeutic
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1885

drugs. Hence, combination treatments are a natural extension of the work presented here.
We also plan to extend the current model to investigate the effect of the hypoxia-inducible
factor α pathway and reactive oxygen species in wound healing and, more particularly,
during treatment with HBOT. A more realistic model of the metabolism that includes aer-
obic and anaerobic respiration may further elucidate how HBOT stimulates the healing of
chronic wounds.
Wounds tend to fail due a multitude of complications (Bauer et al., 2005a), and so
we would like to simulate this using the model we have developed here. We intend to
extend our model to two dimensions in order to more accurately capture the process of re-
epithelialisation and wound closure. We also plan to investigate alternative approaches to
modelling angiogenesis, including a model that focuses on the rate-limiting step of ECM
degradation by proteases.
Our model simulations can provide some indication of when HBOT is unlikely to be
effective. For instance, HBOT will not be effective when used to treat wounds which fail
to heal due to a lack of growth factor production or low chemotactic responsiveness (see
Fig. 7). These results reinforce our hypothesis that it is not how much chemoattractant is
being produced in the wound site, but rather its spatial location that is important.
The results of this paper are of clinical significance. We have shown that for a wound
that fails to heal due to an overabundance of bacteria, intermittent HBOT can accelerate
the rate of healing of a chronic wound but when the treatment is stopped early (at 2 weeks
post injury), healing stalls. Importantly, we have also demonstrated that normobaric oxy-
gen is not a replacement for HBOT and supernormal healing is not an expected outcome.
Our simulations illustrate that HBOT has little benefit for treating normal wounds, and
that exposing a patient to fewer, longer sessions of oxygen is not an appropriate treatment
option.

Acknowledgements

This work was supported by the award of a doctoral scholarship from the Institute of
Health and Biomedical Innovation at Queensland University of Technology (J.A.F.) and
under the Australian Research Council Discovery Project funding scheme (project num-
ber DP0878011) (J.A.F.). This work was carried out while H.M.B. was visiting Queens-
land University of Technology, funded by the Institute of Health and Biomedical Inno-
vation and the School of Mathematical Sciences. Computational resources and services
used in this work were provided by the HPC and Research Support Unit, QUT. The au-
thors would like to acknowledge the very helpful comments of the anonymous referees.

Appendix A

A.1 Methods

In the wound healing process, the cell movement associated with chemotaxis generally
outweighs the movement associated with random motion, leading to advection-dominated
mathematical models of wound healing. The equations in these models must be solved
with care in order to avoid non-physical solution behaviour. We solve the PDEs with a
1886 Flegg et al.

Table 5 List of dimensionless parameters

λ L2 λ L2 e λ λ4 L2 n0 λ L3 n
k1 = D1 c k2 = D
2 0 k3 = c 3 k4 = k5 = D
5 0
c 0 c c0 0 Dc c c0
λ6 L3 n0 λ L2 λ L2 λ9 L 3 γ λ10 γ L2
k6 = Dc k7 = D
7 k8 = D 8e γ k9 = Dc k10 = Dc
c c 0
λ11 L2 n0 λ12 L3 n0 λ13 L3 n0 λ e λ f
k11 = Dc k12 = Dc k13 = Dc k14 = n14L0 k15 = n15L0
0 0
λ L2 λ L2 f λ18 L2 f0 c0 λ c
k16 = 16
Dc k17 = D17 c e 0 k18 = Dc k19 = 19e0
0 κa = Da
Dc
c 0 0
χ cup c
κn = D χen γ κf = D fγ cH = cL = dwn
c 0 c c0 c0

flux limiting finite volume method. For a more detailed review of the numerical solution of
advection-dominated wound healing initial boundary value problems, see Thackham et al.
(2009). We non-dimensionalise Eqs. (1)–(6) by taking c = c0 ĉ, n = n0 n̂, b = b0 b̂ = Ln0 b̂,
a = γ â, f = f0 fˆ, e = e0 ê, x = Lx̂, and t = DLw tˆ. Dropping the hats for notational sim-
2

plicity and denoting dimensionless parameters as per Table 5, we arrive at the dimension-
less equations given in (A.1).
Oxygen, c(x, t):

∂c ∂ 2 c (k1 + k2 e)c  
= 2− − k4 bc + k5 b 1 + αg(t) . (A.1a)
∂t ∂x k3 + c

Chemoattractant, a(x, t):

∂a ∂ 2a k8 H (c − cL )H (cH − c)
= κa 2 − k6 ab − k7 a + . (A.1b)
∂t ∂x 1+e

Capillary tips, n(x, t):

 
∂n ∂ −κn en ∂a
= + a(k9 b + k10 n) − n(k11 n + k12 b). (A.1c)
∂t ∂x (1 + e2 )(1 + a)2 ∂x

Blood vessels, b(x, t):

∂b −κn en ∂a
= + k13 b(k14 e + k15 f − b). (A.1d)
∂t (1 + e )(1 + a) ∂x
2 2

Fibroblasts, f (x, t):

 
∂f ∂ −κf f ∂a k16 f c k17 f 2
= + − . (A.1e)
∂t ∂x (1 + a) ∂x
2 1+c (1 + c)(1 + e)

ECM, e(x, t):

∂e
= k18 f c(k19 c − e). (A.1f)
∂t
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1887

A.2 Boundary and initial conditions

We consider a Cartesian geometry and, for simplicity, take the wound to be one di-
mensional, of length 2L, symmetric about its centre at x = 0 with its wound edge at
x = L. Initially, the wound is devoid of each of the species, and within the healed region
(L ≤ x ≤ 2L) the oxygen, capillary tip, blood vessel, fibroblast, and ECM species take
their value from that of normal, unwounded, tissue (c0 , n0 , b0 , f0 and e0 ), respectively. At
t = 0 on 0 ≤ x < L:

c(x, 0) = a(x, 0) = n(x, 0) = b(x, 0) = f (x, 0) = e(x, 0) = 0,

and at t = 0 on L ≤ x < 2L:

c(L, 0) = c0 ; a(L, 0) = 0; n(L, 0) = n0 ;

b(L, 0) = b0 ; f (L, 0) = f0 ; e(L, 0) = e0 .

Since we have assumed symmetry of the wound about its center, zero flux conditions are
imposed at x = 0:

∂c  ∂a 
 =  = 0. (A.2)
∂x x=0 ∂x x=0
We impose no flux boundary conditions for the chemoattractant, capillary tips and
fibroblasts at x = 2L. The oxygen concentration is taken to satisfy the following equation
at x = 2L:
 
c(2L, t) = c0 1 + αg(t) .

The blood vessel and ECM densities satisfy ODEs and require no boundary condi-
tions.

References

Abidia, A., Laden, G., Kuhan, G., Johnson, B.F., Wilkinson, A.R., Renwick, P.M., Masson, E.A., McCol-
lum, P.T., 2003. The role of hyperbaric oxygen therapy in ischaemic diabetic lower extremity ulcers:
a double-blind randomised-controlled trial. Eur. J. Vasc. Endovasc. Surg. 25, 513–518.
Acker, T., Beck, H., Plate, K.H., 2001. Cell type specific expression of vascular endothelial growth factor
and angiopoietin-1 and -2 suggests an important role of astrocytes in cerebellar vascularization. Mech.
Dev. 108(1–2), 45–57.
Adams, R.H., Alitalo, K., 2007. Molecular regulation of angiogenesis and lymphangiogenesis. Nature Rev.
Mol. Cell Biol. 8, 464–478.
Agren, M.S., Eaglstein, W.H., Ferguson, M.W.J., Harding, K.G., Moore, K., Saarialho-Kere, U.K., Schultz,
G.S., 2000. Causes and effects of the chronic inflammation in venous leg ulcers. Acta Derm. Venereol.
Suppl. 210, 3–17.
Al-Waili, N.S., Butler, G.J., 2006. Effects of hyperbaric oxygen on inflammatory response to wound and
trauma: possible mechanism of action. Sci. World 6, 425–441.
Alessio, M., Gruarin, P., Castagnoli, C., Trombotto, C., Stella, M., 1998. Primary ex vivo culture of ker-
atinocytes isolated from hypertrophic scars as a means of biochemical characterization of CD36. Int.
J. Clin. Lab. Res. 28(1), 47–54.
1888 Flegg et al.

Allen, D.B., Maguire, J.J., Mahdavian, M., Wicke, C., Marcocci, L., Scheuenstuhl, H., Chang, M.,
Le, A.X., Hopf, H.W., Hunt, T.K., 1997. Wound hypoxia and acidosis limit neutrophil bacterial killing
mechanisms. Arch. Surg. 132(9), 991–996.
Anonymous, 1999. American diabetes association: consensus development conference on diabetic foot
wound care. Diabetes Care 22, 1354–1360.
Babul, S., Rhodes, E.C., 2000. The role of hyperbaric oxygen therapy in sports medicine. Sports Med.
30(6), 395–403.
Balding, D., McElwain, D.L.S., 1985. A mathematical model of tumour-induced capillary growth. J. Theor.
Biol. 114(1), 53–73.
Bauer, S.M., Bauer, R.J., Liu, Z.J., Chen, H., Goldstein, L., Velazquez, O.C., 2005a. Vascular en-
dothelial growth factor-C promotes vasculogenesis, angiogenesis, and collagen constriction in three-
dimensional collagen gels. J. Vasc. Surg. 41(4), 699–707.
Bauer, S.M., Bauer, R.J., Velazquez, O.C., 2005b. Angiogenesis, vasculogenesis, and induction of healing
in chronic wounds. Vasc. Endovasc. Surg. 39(4), 293–306.
Boykin, J.V., Baylis, C., 2007. Hyperbaric oxygen therapy mediates increased nitric oxide production
associated with wound healing: a preliminary study. Adv. Skin Wound Care 20(7), 382–389.
Byrne, H.M., Chaplain, M.A.J., 1996. Explicit solutions of a simplified model of capillary sprout growth
during tumour angiogenesis. Appl. Math. Lett. 9(1), 69–74.
Byrne, H.M., Chaplain, M.A.J., Evans, D.L., Hopkinson, I., 2000. Mathematical modelling of angiogenesis
in wound healing: comparison of theory and experiment. J. Theor. Med. 2, 175–197.
Carmeliet, P., 2004. Manipulating angiogenesis in medicine. J. Int. Med. 255, 538–561.
Chang, N., Goodson, W.H., Gottrup, F., Hunt, T.K., 1983. Direct measurement of wound and tissue oxygen
tension in postoperative patients. Ann. Surg. 197(4), 470–478.
Chaplain, M.A.J., Orme, M.E., 1998. Mathematical Modeling of Tumor-Induced Angiogenesis,
Birkhauser, Bosu. Chapter 3.4, pp. 205–240.
Chen, C., Schultz, G.S., Bloch, M., Edwards, P.D., Tebes, S., Mast, B.A., 1999. Molecular and mechanistic
validation of delayed healing rat wounds as a model for human chronic wounds. Wound Repair Regen.
7(6), 486–494.
Chen, S., Yu, C., Cheng, Y., Yu, S., Lo, H., 2007. Effects of hyperbaric oxygen therapy on circulatory
interleukin-8, nitric oxide, and insulin-like growth factors in patients with type 2 diabetes mellitus.
Clin. Biochem. 40, 30–36.
Chen, W.Y.J., Rogers, A.A., 2007. Recent insights into the causes of chronic leg ulceration in venous
diseases and implications on other types of chronic wounds. Wound Repair Regen. 15(4), 434–449.
Cianci, P., 2004. Advances in the treatment of the diabetic foot: is there a role for adjunctive hyperbaric
oxygen. Wound Repair Regen. 12(1), 2–10.
Clark, R.A.F., 1996. In: The Molecular and Cellular Biology of Wound Repair, Wound Repair Overview
and Considerations, pp. 3–35. Plenum, New York. Chapter 1.
Clark, R.A.F., Ghosh, K., Tonnesen, M.G., 2007. Tissue engineering for cutaneous wounds. J. Investig.
Dermatol. 127, 1018–1029.
Cliff, W.J., 1963. Observations on healing tissue: a combined light and electron microscopic investigation.
Philos. Trans. R. Soc. Lond. Ser. B, Biol. Sci. 246(733), 305–325.
Cook, J., 1995. Mathematical models for dermal wound healing: wound contraction and scar formation.
PhD thesis, University of Washington.
David, L.A., Sandor, G.K.B., Evans, A.W., Brown, D.H., 2001. Hyperbaric oxygen therapy and mandibular
osteoradionecrosis: a retrospective study and analysis of treatment outcomes. J. Can. Dent. Assoc. 67,
384.
Dewhirst, M.W., 1994. Determination of local oxygen consumption rates in tumors. Cancer Res. 54(13),
3333–3336.
Diegelmann, R.F., Evans, M.C., 2004. Wound healing: an overview of acute, fibrotic and delayed healing.
Frontiers Biosci. 9, 283–289.
Dimitrijevich, S.D., Paranjape, S., Wilson, J.R., Gracy, R.W., Mills, J.G., 1999. Effect of hyperbaric oxy-
gen on human skin cells in culture and in human dermal and skin equivalents. Wound Repair Regen.
7(1), 53–64.
Dor, Y., Djonov, V., Keshet, E., 2003. Making vascular networks in the adult: branching morphogenesis
without a roadmap. Trends Cell Biol. 13(3), 131–136.
Edelstein, L., 1982. The propagation of fungal colonies: a model for tissue growth. J. Theor. Biol. 98,
671–701.
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1889

Edelstein, L., Segel, L.A., 1983. Growth and metabolism in mycelial fungi. J. Theor. Biol. 104(22), 187–
210.
Edelstein, L., Hadar, Y., Chet, I., Henis, Y., Segel, L.A., 1983. A model for fungal colony growth applied
to sclerotium reolfsii. J. Gen. Microbiol. 129, 1873–1881.
Eginton, M.T., Brown, K.R., Seabrook, G.R., Towne, J.B., Cambria, R.A., 2003. A prospective randomized
evaluation of negative-pressure wound dressings for diabetic foot wounds. J. Ann. Vasc. Surg. 17(6),
645–649.
Enoch, S., Price, P., 2004. Cellular, molecular and biochemical differences in the pathophysiology of heal-
ing between acute wounds, chronic wounds and wounds in the aged. World Wide Wounds. Available
from http://www.worldwidewounds.com//2004/august/Enoch/Pathophysiology-Of-Healing.html.
Faglia, E., Favales, F., Aldeghi, A., Calia, P., Quarantiello, A., Oriani, G., Michael, M., Campagnoli, P.,
Morabito, A., 1996. Adjunctive systemic hyperbaric oxygen therapy in treatment of severe prevalently
ischemic diabetic foot ulcers. Diabetes Care 19(12), 1338–1343.
Fife, C.E., Buyukcakir, C., Otto, G.H., Sheffield, P.J., Warriner, R.A., Love, T.L., Mader, J., 2002. The
predictive value of transcutaneous oxygen tension measurement in diabetic lower extremity ulcers
treated with hyperbaric oxygen therapy: a retrospective analysis of 1144 patients. Wound Repair Re-
gen. 10(4), 198–207.
Flegg, J.A., McElwain, D.L.S., Byrne, H.M., Turner, I.W., 2009. A three species model to simulate appli-
cation of hyperbaric oxygen therapy to chronic wounds. PLoS Comput. Biol. 5(7), e1000451.
Fries, R.B., Wallace, W.A., Roy, S., Kuppusamy, P., Bergdall, V., Gordillo, G.M., Melvin, W.S., Sen,
C.K., 2005. Dermal excisional wound healing in pigs following treatment with topically applied pure
oxygen. Mutat. Res.-Fundam. Mol. Mech. Mutagenes. 579(1–2), 172–181.
Gaffney, E.A., Pugh, K., Maini, P.K., Arnold, F., 2002. Investigating a simple model of cutaneous wound
healing angiogenesis. J. Math. Biol. 45(4), 337–374.
Gajendrareddy, P.K., Sen, C.K., Horan, M.P., Marucha, P.T., 2005. Hyperbaric oxygen therapy ameliorates
stress-impaired dermal wound healing. Brain Behav. Immun. 19(3), 217–222.
Gallagher, K.A., Goldstein, L.J., Thom, S.R., Velazquez, O.C., 2006. Hyperbaric oxygen and bone
marrow-derived endothelial progenitor cells in diabetic wound healing. Vascular 14(6), 328–37.
Gill, A.L., Bell, C.N.A., 2004. Hyperbaric oxygen: its uses, mechanisms of action and outcomes. Q. J.
Med. 97, 385–395.
Gordillo, G.M., Sen, C.K., 2003. Revisiting the essential role of oxygen in wound healing. Am. J. Surg.
186, 259–263.
Gordillo, G.M., Roy, S., Khanna, S., Schlanger, R., Khanderlwal, S., Phillips, G., Sen, C., 2008. Top-
ical oxygen therapy induces vascular endothelial growth factor expression and improves closure of
clinically presented chronic wounds. Clin. Exp. Pharmacol. Physiol. 35, 957–964.
Gottrup, F., 2004. Oxygen in wound healing and infection. World J. Surg. 28(3), 312–315.
Grey, J.E., Harding, K.G. (Eds.), 2006. ABC of Wound Healing. Blackwell, Oxford.
Hammarlund, C., Sundberg, T., 1994. Hyperbaric oxygen reduced size of chronic leg ulcers: a randomised
double-blind study. Plast. Reconstr. Surg. 93(4), 829–834.
Harmey, J.H., Dimitriadis, E., Kay, E., Redmond, H.P., Bouchier-Hayes, D., 1998. Regulation of
macrophage production of vascular endothelial growth factor (VEGF) by hypoxia and transforming
growth factor β − 1. Ann. Surg. Oncol. 5(3), 271–278.
Hehenberger, K., Brismar, K., Lind, F., Kratz, G., 1997. Dose-dependent hyperbaric oxygen stimulation
of human fibroblast proliferation. Wound Repair Regen. 5(2), 147–150.
Hollander, D.A., Hakimi, M.Y., Hartmann, A., Wilhelm, K., Windolf, J., 2000. The influence of hyperbaric
oxygenation (HBO) on proliferation and differentiation of human keratinocyte cultures in vitro. Cell
Tissue Bank. 1, 261–269.
Hopf, H.W., Gibson, J.J., Angeles, A.P., Constant, J.S., Feng, J.J., Rollins, M.D., Hussain, M.Z., Hunt,
T.K., 2005. Hyperoxia and angiogenesis. Wound Repair Regen. 13(6), 558–564.
Hopf, H.W., Ueno, C., Aslam, R., Burnand, K., Fife, C., Grant, L., Holloway, A., Iafrati, M.D., Mani,
R., Misare, B., Rosen, N., Shapshak, D., Slade, J.B., West, J., Barbul, A., 2006. Guidelines for the
treatment of arterial insufficiency ulcers. Wound Repair Regen. 14(6), 693–710.
Hunt, T.K., Zederfeldt, B., Goldstick, T.K., 1969. Oxygen and healing. Am. J. Surg. 118(4), 521–525.
Hunt, T.K., Twomey, P., Zederfeldt, B., Dunphy, J.E., 1967. Respiratory gas tensions and pH in healing
wounds. Am. J. Surg. 114(2), 302–7.
Ichioka, S., Ando, T., Shibata, M., Sekiya, N., Nakatsuka, T., 2008. Oxygen consumption of keloids and
hypertrophic scars. Ann. Plast. Surg. 60(2), 194.
1890 Flegg et al.

James, P.B., Scott, B., Allen, M.W., 1993. Hyperbaric oxygen therapy in sports injuries. Physiotherapy
79(8), 571–572.
Keller, E.F., Segel, L.A., 1971. Traveling bands of chemotactic bacteria: a theoretical analysis. J. Theor.
Biol. 30, 235–248.
Kessler, L., Bilbault, P., Ortega, F., Grasso, C., Passemard, R., Stephan, D., Pinget, M., Schneider, F.,
2003. Hyperbaric oxygenation accelerates the healing rate of nonischemic chronic diabetic foot ulcers.
Diabetes Care 26(8), 2378–2382.
Knighton, D.R., Silver, I.A., Hunt, T.K., 1981. Regulation of wound-healing angiogenesis—effect of oxy-
gen gradients and inspired oxygen concentration. Surgery 90, 262–270.
Leach, R.M., Rees, P.J., Wilmshurst, P., 1998. ABC of oxygen: hyperbaric oxygen therapy. Br. Med. J.
317, 1140–1143.
Lerman, O.Z., Galiano, R.D., Armour, M., Levine, J.P., Gurtner, G.C., 2003. Cellular dysfunction in the
diabetic fibroblast. Am. J. Pathol. 162(1), 303–312.
Levine, H.A., Sleeman, B.D., Nilsen-Hamilton, M., 2000. A mathematical model for the roles of pericytes
and macrophages in the initiation of angiogenesis. I. The role of protease inhibitors in preventing
angiogenesis. Math. Biosci. 168(1), 77–115.
Levine, H.A., Pamuk, S., Sleeman, B.D., Nilsen-Hamilton, M., 2001. Mathematical modeling of capil-
lary formation and development in tumor angiogenesis: penetration into the stroma. Bull. Math. Biol.
63(5), 801–863.
Levine, H.A., Pamuk, S., Sleeman, B.D., Nilsen-Hamilton, M., 2002. Mathematical modelling of tumour
angiogenesis and the action of angiostatin as a protease inhibitor. J. Theor. Med. 4(2), 133–145.
Lin, H., Chi, S., Perng, W., Wu, C., Lin, Z., Huang, K., 2008. Hyperbaric oxygen attenuates cell growth in
skin fibroblasts cultured in a high-glucose medium. Wound Repair Regen. 16(4), 513–519.
Mace, K.A., Yu, D.H., Paydar, K.Z., Boudreau, N., Young, D.M., 2007. Sustained expression of Hif-1alpha
in the diabetic environment promotes angiogenesis and cutaneous wound repair. Wound Repair Regen.
15(5), 636–645.
Mathieu, D., 2002. Hyperbaric Surgery, Hyperbaric Oxygen Therapy in the Management of Non-healing
Wounds, pp. 317–339. Best Publishing Company. Chapter 12.
Moore, K., Ruge, F., Harding, K.G., 1997. T lymphocytes and the lack of activated macrophages in wound
margin biopsies from chronic leg ulcers. Br. J. Dermatol. 137, 188–194.
Murray, J.D., Oster, G.F., Harris, A.K., 1983. A mechanical model for mesenchymal morphogenesis.
J. Math. Biol. 17(1), 125–129.
Murray, J.D., Maini, P.K., Tranquillo, R.T., 1988. Mechanochemical models for generating biological
pattern and form in development. Phys. Rep. 171(2), 59–84.
Niklas, A., Brock, D., Schober, R., Schulz, D., Schneider, A., 2004. Continuous measurements of cerebral
tissue oxygen pressure during hyperbaric oxygenation–HBO effects on brain edema and necrosis after
severe brain trauma in rabbits. J. Neurol. Sci. 219(1–2), 77–82.
Oberringer, M., Meins, C., Bubel, M., Pohlemann, T., 2007. A new in vitro wound model based on the
co-culture of human dermal microvascular endothelial cells and human dermal fibroblasts. Biol. Cell.
99(4), 197–207.
Olaso, E., Labrador, J.P., Wang, L.H., Ikeda, K., Eng, F.J., Klein, R., Lovett, D.H., Lin, H.C., Fried-
man, S.L., 2002. Discoidin domain receptor 2 regulates fibroblast proliferation and migration through
the extracellular matrix in association with transcriptional activation of matrix metalloproteinase-2.
J. Biol. Chem. 277(5), 3606–3613.
Olsen, L., Sherratt, J.A., Maini, P.K., 1995. A mechanochemical model for adult dermal wound contraction
and the permanence of the contracted tissue displacement profile. J. Theor. Biol. 177, 113–128.
Olsen, L., Sherratt, J.A., Maini, P.K., Arnold, F., 1997. A mathematical model for the capillary endothelial
cell-extracellular matrix interactions in wound-healing angiogenesis. Math. Med. Biol. 14(4), 261–
281.
Olsen, L., Maini, P.K., Sherratt, J.A., 1998. Spatially varying equilibria of mechanical models: application
to dermal wound contraction. Math. Biosci. 147(1), 113–129.
Oyibo, S.O., Jude, E.B., Tarawneh, I., Nguyem, H.C., Armstrong, D.G., Harkless, L.B., Boulton, A.J.M.,
2001. The effects of ulcer size and site, patient’s age, sex and type and duration of diabetes on the
outcome of diabetic foot ulcers. Diabetic Med. 18(2), 133–138.
Panovska, J., Byrne, H.M., Maini, P.K., 2008. A theoretical study of the response of vascular tumours to
different types of chemotherapy. Math. Comput. Model. 47, 560–579.
Pettet, G.J., Byrne, H.M., McElwain, D.L.S., Norbury, J., 1996a. A model of wound-healing angiogenesis
in soft tissue. Math. Biosci. 136(1), 35–63.
Mathematical Model of Hyperbaric Oxygen Therapy Applied 1891

Pettet, G.J., Chaplain, M.A.J., McElwain, D.L.S., Byrne, H.M., 1996b. On the role of angiogenesis in
wound healing. Proc. R. Soc. Lond. B 263(1376), 1487–1493.
Piantadosi, C.A., 1999. Physiology of hyperbaric hyperoxia. Respir. Care Clin. North Am. 5(1), 7–19.
Polverini, P.J., 1995. The pathophysiology of angiogenesis. Critical Rev. Oral Biol. Med. 6(3), 230–247.
Polverini, P.J., 2002. Angiogenesis in health and disease: insights into basic mechanisms and therapeutic
opportunities. J. Dent. Educ. 66(8), 962–975.
Raa, A., Stansberg, C., Steen, V.M., Bjerkvig, R., Reed, R.K., Stuhr, L.E.B., 2007. Hyperoxia retards
growth and induces apoptosis and loss of glands and blood vessels in DMBA-induced rat mammary
tumors. BMC Cancer 7, 23.
Said, H.K., Hijjawi, J., Roy, N., Mogford, J., Mustoe, T., 2005. Transdermal sustained-delivery oxygen
improves epithelial healing in a rabbit ear wound model. Arch. Surg. 140, 998–1004.
Schugart, R.C., Friedman, A., Zhao, R., Sen, C.K., 2008. Wound angiogenesis as a function of tissue
oxygen tension: a mathematical model. Proc. Natl. Acad. Sci. 105(7), 2628.
Sen, C.K., Khanna, S., Gordillo, G., Bagchi, D., Bagchi, M., Roy, S., 2002. Oxygen, oxidants, and antiox-
idants in wound healing. Ann. N.Y. Acad. Sci. 957, 239–249.
Sen, C.K., Gordillo, G.M., Roy, S., Kirsner, R., Lambert, L., Hunt, T.K., Gottrup, F., Gurtner, G.C., Lon-
gaker, M.T., 2009. Human skin wounds: a major and snowballing threat to public health and the
economy. Wound Repair Regen. 17(6), 763–771.
Serini, G., Ambrosi, D., Giraudo, E., Gamba, A., Preziosi, L., Bussolino, F., 2003. Modeling the early
stages of vascular network assembly. EMBO J. 22, 1771–1779.
Sheffield, P.J., 1985. Tissue oxygen measurements with respect to soft tissue wound healing with normo-
baric and hyperbaric oxygen. HBO Rev. 6(1), 18–46.
Silverman, W.A., 2004. A cautionary tale about supplemental oxygen: the albatross of neonatal medicine.
Pediatrics 113(2), 394–396.
Simons, M., 2005. Angiogenesis, arteriogenesis, and diabetes paradigm reassessed? J. Am. College Car-
diol. 46(5), 835–837.
Singer, A.J., Clark, R.A.F., 1999. Cutaneous Wound Healing. New Engl. J. Med. 341(10), 738–746.
Slovis, N., 2006. Veterinary hyperbaric medicine society: review of hyperbaric medicine. On line.
Sola, A., Saldeño, Y.P., Favareto, V., 2008. Clinical practices in neonatal oxygenation: where have we
failed? What can we do? J. Perinatol. 28(1), S28–34.
Sun, S., Wheeler, M.F., Obeyesekere, M., Patrick, C., 2005. Nonlinear behaviors of capillary formation in
a deterministic angiogenesis model. Nonlinear Anal. 63, 2273–2246.
Tandara, A.A., Mustoe, T.A., 2004. Oxygen in wound healing—more than a nutrient. World J. Surg. 28,
294–300.
Thackham, J.A., McElwain, D.L.S., Long, B., 2008. The use of hyperbaric oxygen therapy to treat chronic
wounds: a review. Wound Repair Regen. 16(3), 321–330.
Thackham, J.A., McElwain, D.L.S., Turner, I.W., 2009. Computational approaches to solving equations
arising from wound healing. Bull. Math. Biol. 71, 211–246.
Thom, S.R., 2009. Oxidative stress is fundamental to hyperbaric oxygen therapy. J. Appl. Phys. 106, 988–
995.
Tompach, P.C., Lew, D., Stoll, J.L., 1997. Cell response to hyperbaric oxygen treatment. Int. J. Oral.
Maxillofac. Surg. 26(2), 82–86.
Trabold, O., Wagner, S., Wicke, H., Scheuenstuhl, C., Hussain, M.Z., Rosen, N., Seremetiev, A., Becker,
H.D., Hunt, T.K., 2003. Lactate and oxygen constitute a fundamental regulatory mechanism in wound
healing. Wound Repair Regen. 11(6), 504–509.
Tranquillo, R.T., Murray, J.D., 1992. Continuum model of fibroblast-driven wound contraction:
inflammation-mediation. J. Theor. Biol. 158, 135–172.
Williams, R.L., 1997. Hyperbaric oxygen therapy and the diabetic foot. J. Am. Podiatr. Med. Assoc. 87(6),
279–292.
Woo, T.C., Joseph, D., Oxer, H., 1997. Hyperbaric oxygen treatment for radiation proctitis. Int. J. Radiat.
Oncol. Biol. Phys. 38(3), 619–22.
Xue, C., Friedman, A., Sen, C.K., 2009. A mathematical model of ischemic cutaneous wounds. Proc. Nat.
Acad. Sci. 106(39), 16782.