Digital three-dimensional modelling of the male pelvis and bicycle seats: impact of rider position and seat design on potential penile hypoxia and erectile dysfunction


John Gemery, Interventional Radiology, Dartmouth-Hitchcock Medical Center, One Medical Center Drive, Lebanon NH 03756, USA. e-mail:



To digitally model (three-dimensional, 3D) the course of the pudendal arteries relative to the bony pelvis in the adult male, and to identify sites of compression with different bicycle riding positions as a potential cause of penile hypoxia and erectile dysfunction.


3D models were made from computed tomography scans of one adult male pelvis (a healthy volunteer) and three bicycle seats. Models were correlated with lateral radiographs of a seated rider to determine potential vascular compression between the bony pelvis and seats at different angles of rider positioning.


Pelvis/seat models suggest that the most likely site of compression of the internal pudendal artery is immediately below the pubic symphysis, especially with the rider leaning forward. For an upright rider, the internal pudendal arteries do not appear to be compressed between the seat and the bony pelvis. Leaning partly forward with arms extended, the seat/symphysis areas were reduced to 73 mm2 with standard seat and 259 mm2 with a grooved seat. Leaning fully forward, the seat/symphysis areas decreased (no space with standard seat; 51 mm2 with a grooved seat) and both the ischial tuberosities and the pubic symphysis might be in contact with the seat.


A grooved seat allows better preservation of the seat/symphysis space than a standard seat, but the rider’s position is more important for preserving the seat-symphysis space (and reducing compression) than is seat design alone. Any factors which influence the seat-symphysis space (including an individual’s anatomy, seat design and rider position) can increase the potential for penile hypoxia and erectile dysfunction/perineal numbness.


erectile dysfunction


3D, two, three-dimensional


Medical Metrx Solutions.


Bicycle riding has been linked to an increased risk of erectile dysfunction (ED) and abnormal perineal sensations in male riders, which are postulated to result from compression of the internal pudendal arteries and/or the pudendal nerves [1–8]. The artery and nerve typically course together on each side of the pelvis, and compression of one would be expected to lead to compression of the other. Seat manufacturers have responded with a variety of seat designs intended to reduce the potential for pudendal artery and nerve compression. The different seat designs have been shown to have different effects on penile de-oxygenation during riding [9]. The objective of the present study was to create digital three-dimensional (3D) models of a pelvis, pudendal arteries and bicycle seats to evaluate potential sites of compression of the vessels. We hypothesized that the type of seat in conjunction with the rider’s position differentially affects the orientation and compression of the pudendal arteries.


The authors created a digital male pelvis model (in 3D) and combined it with digital models of sample bicycle seat designs (in 3D), in positions that might be adopted during riding, to evaluate for potential compression of the internal pudendal arteries between the bony pelvis and bicycle seat. The digital pelvis model data were derived from contrast-enhanced CT of an adult male volunteer. The subject was scanned with a four-detector scanner (Light Speed, General Electric, Milwaukee, WI, USA), with 2.5 mm slices, using a standard algorithm and 140 mL of i.v. iodinated contrast medium (Omnipaque 350, Amersham Health Inc., Princeton NJ, USA). The data were reformatted to 1.25 mm slices, with 50% overlap, transferred to an optical disk and then sent to a digital modelling company (Medical Metrx Solutions, MMS, West Lebanon, NH, USA) to create a digital 3D pelvis model including the bony pelvis, the iliac arteries, the internal pudendal arteries from origin to below the pubic symphysis, and the proximal portions of other internal iliac artery branches. Creating the 3D model components was based on identifying each on the original 2D CT scan images.

The margins of the bones of the pelvis were readily evident and easily modelled. One of the authors (J.M.G.) worked directly with MMS at their workstation to identify the courses of the vessels to be modelled. These included the common and external iliac arteries bilaterally, the proximal portions of the posterior divisions of the internal iliac arteries, and the proximal anterior divisions of the internal iliac arteries. The internal pudendal arteries were identified as those vessels originating from the anterior division of the internal iliac arteries and coursing to the base of the penis on each side. The modelling software (Preview 2.0.2, MMS) is used commercially for modelling abdominal aortic aneurysms for planning endovascular repair.

Three bicycle seat designs were chosen to represent the range of designs available (Fig. 1); these included a tensioned leather bicycle seat (right; Brooks Professional, Brooks Saddle Co., Birmingham, UK); a narrow racing seat (centre; Flite, Selle Italia, Rossano Veneto, Italy); and a design with a central groove intended to reduce perineal pressure (left; Body Geometry Comp Saddle, Specialized, Morgan Hill, CA, USA).

Figure 1.

Bicycle seat designs used in the modelling: A, a tensioned leather seat (right; Brooks Professional); a narrow racing seat (centre; Flite, Selle Italia); and a design with a central groove intended to reduce perineal pressure (left; Body Geometry Comp Saddle, Specialized).

A wax mould of each seat was created and the moulds scanned to produce the digital information for the digital seat models. The seats were felt to contain sufficient metal that the CT artefacts generated by the metal would preclude accurate modelling of the contours if they were scanned directly. A disposable aluminium pan was placed in a larger container of boiling water and ≈ 5 kg of food-grade wax (Northland Wax, Conros Corp., Detroit, MI, USA) was placed in the pan and melted. Each seat was wrapped with a single layer of aluminium foil, hand-smoothed to conform as closely as possible to the curvatures of each seat design. The foil was intended to minimize damage to the seats and to facilitate later removal of the seat from the wax. The foil covered seat was then inverted and placed in the molten wax, which was removed from the heat and allowed to cool. When the wax had cooled sufficiently to retain the shape, the seat was removed from the wax and the resulting wax block containing the imprint of the seat was compared to the seat for accuracy. The wax block was not removed from the aluminium pan.

The aluminium pan containing the wax block with the seat imprint was placed on the CT gantry and helically scanned perpendicular to the long axis of the seat. The technique used was helical 2.5-mm thick scans. The initial seat phantom (the central groove design) was scanned on a single-detector scanner at 3 mm slice thickness, with 50% overlap, and that information was stored on the optical disk with no reformatting. The other two seat phantoms were scanned on the multidetector scanner, at 2.5 mm slice thickness with 50% overlap, and that information was also stored on the optical disk with no reformatting. The same modelling process was repeated for each subsequent seat. The digital data obtained were processed by MMS to create digital models of the upper surfaces of each bicycle seat that were in contact with the rider.

To obtain information on the correct orientation of the bony pelvis relative to the bicycle seat, lateral plain film radiographs were obtained of an adult male volunteer sitting on a bicycle in positions intended to model the extremes of possible seating positions. These ranged from fully upright (Fig. 2A) through semi-upright with arms extended straight to the handlebars (Fig. 2B) to maximum forward lean, as when using an aerodynamic bar (Fig. 2C). The aerodynamic bar brings the elbows and forearms together and to the level of the centre of the handle bars, and is intended to produce an aerodynamic profile, minimizing wind resistance. The rider and bicycle were stabilized in a near-upright position against the radiography film holder. This positioning both stabilized the rider in as representative a position as possible, and minimized the radiation dose to the subject by minimizing the air gap between subject and film. The images were obtained on a bicycle (Klein Quantum, Waterloo, WI, USA) with a 57-cm frame with a Body Geometry S Works seat (Specialized) and an Airstryke 2000 aerobar (Profile, Carson CA, USA). Both the pelvic CT and the lateral plain films were obtained in accordance with Institutional Review Board guidelines for the protection of human subjects.

Figure 2.

(A–C) Lateral photographs of a seated rider with corresponding photographs used to determine the appropriate positioning for the seat and pelvis models.

A line drawn parallel to the top of the seat and a second line drawn at a tangent to the undersurfaces of the pubic symphysis and ischial tuberosities will intersect at the point of contact between the seat and ischial tuberosities. The angle formed by these two lines provides an estimate of the available space between the undersurface of the pubic symphysis and the top of the seat. This angle was noted on the films obtained for each rider position.

The seat/pelvis angles derived from these lateral images were used to create final models combining the digital seat models with the digital pelvis model. In each case, the angle between the seat models and the pelvis model was set and the seat then ‘raised’ to the inferior aspect of the ischial tuberosities. The seat/pelvis angles chosen to best represent the potential angles encountered in actual riding were 0 and 20°, corresponding to full forward lean with the aerodynamic bars and partial forward lean with arms extended, respectively. In each final seat/pelvis model, the different seat designs can be viewed separately relative to the pelvis (Fig. 3). These models were evaluated with Preview 2.0.2 software (MMS); the relationship of each seat design to the pelvis was noted at both 0 and 20°, and illustrative images saved.

Figure 3.

Frontal views with pelvis positions corresponding to: A, a rider in a partial forward lean with arms extended; and B, a rider in a full forward lean as when using aerodynamic bars. The brown seat has a central groove; the pink seat is a flat racing style.

As a final step, we used a semiquantitative analysis to estimate the available free space between the seat top surface and the undersurface of the pubic symphysis, at 0 and 20° using two different seat designs (Fig. 4). Using the MMS Preview software, one author (J.M.G.) identified points approximating the margins of the spaces between the bike seat and the pubic symphysis (Fig. 4). The MMS software was then used to measure the outlines of those spaces, and basic trigonometry to calculate the areas.

Figure 4.

A semi-quantitative evaluation of the available space between the seat and the bony pelvis. (a) with the rider in the partial forward lean position, with a (pink) standard seat, and geometry measured (the geometry is for reference and not to scale); and b) a rider in the partial forward lean position, with a (brown) grooved seat and the geometry measured (not to scale).


The bony pelvis and proximal femurs are easily visible on CT and easily definable for modelling. The large vessels and many of the smaller vessels of the pelvis are also clear. The common, external and proximal internal iliac arteries were included in the model for reference, as were the proximal common femoral arteries. The internal pudendal arteries were identified and modelled bilaterally (Fig. 3).

The wax moulds of the seats all conformed to the seats themselves, with only minimal differences caused by small wrinkles in the covering foil, that were barely detectable on CT and were not included in the seat models.

A review of the lateral X-ray images of the volunteer on the bicycle showed a decreasing angle, and hence space, between the upper surface of the seat and undersurface of the pubic symphysis with progressive forward lean by the rider.

The 3D pelvis and bicycle seat models, which provide views from the anterior, posterior or any angle the investigator wishes, also showed the decreased gap between the top of seat and undersurface of the pubic symphysis in lateral projection as seen on the lateral plain films (Fig. 3). The anterior and posterior views in particular showed better preservation of the seat/symphysis space in seat designs incorporating a central groove.

The pelvis/seat models suggested that the most likely site of compression of the internal pudendal arteries is immediately below the pubic symphysis, especially with the rider leaning fully forward, as when using aerodynamic bars (Fig. 2A–C). The course of the pudendal artery appeared to be remote from the seat surfaces, except in the region between the seat and pubic symphysis. We found no other likely sites of compression of the pudendal arteries between any of the seat designs and the bony pelvis.

With partial forward lean and the arms extended (20°), the estimated available spaces between the seat and symphysis were 73 mm2 with the narrow racing seat and 259 mm2 with the grooved seat (Fig. 2B). When leaning fully forward (0°) the seat/symphysis areas decreased, with potentially both the pubic symphysis and the ischial tuberosities in contact with the seat. The estimated available spaces were markedly less, i.e. no space with the narrow seat, and 51 mm2 with the grooved seat (Fig. 2C).


Salimpour et al.[5] reported the results of a survey of bicyclists and runners, which concluded that the risk of sexual and urinary tract dysfunction was directly related to bicycling exposure. Schwarzer et al.[6], at the AUA meeting in 1999, reported similar results from a survey of German cyclists and swimmers. In their study, the ED rate among cyclists was 4%, vs 2% in swimmers. Schrader et al.[10] reported an inverse relationship between weekly cycling time and measures of nocturnal erectile quality in police bicycle-patrol officers. Marceau et al.[8], drawing on data from the Massachusetts Male Aging Study, also suggested a link between ED and bicycling for >3 h/week.

Findings such as these and others have led to increased public awareness and concern, with articles on perineal symptoms and ED related to bicycling appearing in the lay press and enthusiast magazines [11–14]. Compression of the internal pudendal artery and pudendal nerve were postulated to be the causative injury leading to perineal symptoms and ED, but whether injury to one, the other or both is necessary for symptoms to occur remains uncertain, as there is evidence for injury to both.

Andersen and Bovim [4] reported the results of a questionnaire returned by 169 people (160 men, 9 women) participating in an amateur non-stop 540 km bicycle race. They noted that ED was reported by 21 of 160 men (13%), lasting >1 week in 11 men and >1 month in three. While this conceivably could have been due to arterial injury, 33 of the 160 men reported genital numbness in the distribution of the bilateral pudendal or cavernosal nerves, lasting a day or more in 12%, and lasting 1–16 weeks in 6%[4].

Ricchiuti et al.[7] reported objective evidence of pudendal nerve damage in a cyclist; they reported abnormal nerve conduction studies, ‘most consistent with chronic, moderate, pudendal nerve (or possibly sacral root) injuries’, in a patient with a history of dorsal penile numbness related to bicycling. The numbness in this man resolved after riding, and the ‘mild transient ED’ also resolved with ‘cessation of the avid bicycling’.

Evidence in favour of arterial injury includes the finding that sitting on a bicycle seat can decrease blood flow to the penis. Kirstein et al.[15] reported that 20 normal volunteers with no ED or ejaculatory dysfunction had a decrease in penile blood pressure relative to ankle and brachial pressures with 5 min of sitting on an unpadded bicycle seat. No mention was made of the type of seat used, or the positioning of the volunteers, but as this study was published in 1982, the seat used would have been unlikely to have had a central cut-out or groove, as such seats were only introduced to the general cycling market after that date.

Penile hypoxia secondary to decreased blood flow might be injurious, and has been postulated to promote trabecular connective tissue synthesis, which could impede the achievement of erection [16]. Sommer et al.[17] measured penile transcutaneous oxygen pressure and showed, in 46 healthy men, a decrease in mean transcutaneous penile oxygen pressure at the glans from 60.5  (+/−8.1) mmHg, standing before cycling, to 17.9 (+/−3.9) mmHg, after 15 min cycling while seated. Schwarzer et al.[9] also reported penile de-oxygenation with bicycle riding, but further showed that the degree of penile de-oxygenation varied with seat design. Nayal et al. reported a similar decline in oxygenation between standing and seated cycling [18].

The precise anatomical site where compression might occur, leading to hypoxia, has not been established. The area between the seat and pubic symphysis has been proposed as the site of compression [1,2]. Other authors have argued that the compression occurs along the medial aspect of the ischial ramus, instead, within Alcock’s canal [3]. Ricchiuti et al.[7] suggested three possible points of compression, ‘inside the pudendal (Alcock’s) canal; between the sacrospinal and sacral tuberous ligaments; and against the pubic arc, as the body is tilted forward on the bicycle seat’. The present model suggested that the area below the pubic symphysis is the point of greatest risk, but does not exclude the possibility of compression within the pudendal canal. However, when viewed in posterior projection the present model suggests that the internal pudendal vessels are sufficiently remote from the seat (at least in the seat designs modelled) to not be trapped between the seat and the forward portion of the ischial tuberosities.

Observations from our model agree with the findings of Schwarzer et al.[9], who reported decreases in penile transcutaneous oxygenation between subjects standing at rest and those cycling using a stationary bicycle. They noted an 82.4% decrease in initial oxygen pressure in subjects seated on a narrow, heavily padded seat; a 72.4% decrease when subjects were seated on a ‘body geometry’ seat (incorporating a central groove); and a 20.3% decrease with subjects sitting on a seat with no ‘seat nose’.

Our model would predict that, for a given degree of forward rider lean, a seat with a central notch would create less potential compression of the pudendal artery than a seat with padding below the pubic symphysis, in accordance with these reported oxygen pressure measurements. Schwarzer et al.[9] reported that a seat with no nose, i.e. no material beneath the pubic symphysis, produced the least de-oxygenation of all the seat designs they tested, again agreeing with what our model would predict. Lowe et al.[19] directly measured seat pressure and found that a sport/racing seat design generated twice the perineal pressure of a seat with no protruding nose. These findings, in combination, suggest that the region below the pubic symphysis is the main determinant of the restriction of penile blood flow, rather than the other areas proposed by Ricchiuti et al., particularly the pudendal (Alcock’s) canal. If correct, then the question of whether symptoms are due to compression of arteries, nerves or both might be moot; both arteries and nerves both pass below the pubic symphysis in most individuals, so strategies that reduce compression of one should also reduce the risk to the other.

Strategies to reduce potential compression should include both positioning and seat design. For any given degree of forward lean, the seats with a central groove offer more potential space between the seat and pubic symphysis than do narrow racing-style seats. However, we showed that this advantage can be lost with changes in the rider’s position. For example, our model estimated an available space of 52 mm2 for a rider leaning fully forward on a grooved seat, which is less than the 73 mm2 available for the same rider sitting more upright on a narrow racing seat. Rider positioning might thus be the prime determinant of available space.

Studies such as that by Schwarzer et al.[9] have not addressed the effect of rider position on potential compression, but focused only on seat design. Our finding that the rider’s position is the main determinant of available space suggests several simple protective strategies for riders: minimizing the amount of time spent riding in a full forward lean, as when using aerodynamic bars; avoiding an upward tilt of the seat nose; and avoiding adding bulk to the seat nose, such as additional padding. These strategies are applicable not just to bicycle riding, but also to seats used in stationary exercise bicycles.

There are some limitations and assumptions made with our model. The seats were assumed to be level with the ground in all models. Positioning of the seat, namely upward tilt of the nose of the seat, would decrease the available space between seat and symphysis in the same manner as forward rider lean. The height of the seat relative to the handlebars has also been assumed to be constant. Raising the seat relative to handlebars would induce forward lean by the rider with greater forward tilt of the pelvis, again decreasing available space for the arteries and nerves. This would suggest that the rider would be less likely to suffer compression by achieving the correct leg extension with a larger-frame bicycle, rather than by raising the seat relative to the handlebars. The seats were also assumed not to be deformable, in that seat advancement relative to the pelvis was assumed to stop when the pelvis model’s ischial tuberosities contacted the seat model. If the seat were padded or soft, such that the ischial tuberosities could sink into the seat, then the space available between seat and symphysis at each angle of forward lean would again decrease.

At present, the model is based on the anatomy of one individual. There are variations in pelvic vessel branching patterns between individuals, and the vessel course in this case might conceivably not be representative of all members of the general population [20,21]. However, other than in rare instances, any vessel supplying the penis would have to pass beneath the pubic symphysis to reach the base of the penis, so the findings from the present model should be valid for much of the population. Additional points of potential compression might be uncovered if further pelves are studied and modelled.

The 3D models of the individual and seats were not scanned with the individual on the seat, but rather the model and the seats were scanned separately, as described. The digital modelling company then combined the 3D models of rider and seat. There might be an inherent human error, in that the distance between rider and seat depended partly on a human co-registering the two 3D models. The semiquantitative measurements only provide useful introspective data on the relative differences in seat/symphysis space for the present subject. These should not be interpreted as spaces that would be available to any rider using these seat designs.

The present model of pelvic anatomy and seat position/lean might explain why ED occurs in some bicycle riders but not all. Why, after the 540 km bicycle race reported by Andersen and Bovim [4], should any rider remain free of symptoms, if all are at risk? Possible explanations include a limited time with full weight bearing on the seat, despite the length of the race, and under-reporting of a sensitive topic. ED and atypical perineal sensations seem to improve with time in some individuals after stopping bicycling, suggesting that permanent injury to the artery and/or nerve might require repeated or chronic insults, with no time allowed for healing [4,7]. Another possibility would be that there is some anatomical variation present in some of the population that puts those individuals at greater risk of injury when bicycling.

In conclusion, the site of compression of the internal pudendal arteries and pudendal nerves between bicycle seats and the bony pelvis remains in question. Data from our model strongly support the hypothesis that the compression occurs between the top of the forward portion of the bicycle seats and the undersurface of the pubic symphysis, and is associated with the rider’s position. Seat designs with a central groove should reduce the risk of compression relative to narrow seats, but only at equivalent degrees of forward lean by the rider. The advantage of a grooved seat is lost as a rider leans further forward, pivoting on the ischial tuberosities and closing the space between the top of the forward portion of seat and the undersurface of the pubic symphysis. Our work suggests that the rider’s position has a greater role than seat design in potential compression.


We would like to thank Dennis D. O’Connor (formerly of MMS, now of DHMC) for extensive modelling work.


Source of funding: the work was supported in part by a grant from The Hitchcock Foundation.