Josie Faulkner

Ndiaye et al. report a head-to-head comparison of a commercial artificial intelligence radiology software (AI) and veterinary radiologists interpreting canine and feline radiographs (1). The conclusion states that the AI “performs almost as well as the best veterinary radiologist in all settings of descriptive radiographic findings” (1). The AI showed high specificity (ability to correctly identify normal findings) but lower sensitivity (ability to correctly detect abnormal findings) than radiologists (1), leading the authors to suggest that its strength lies in confirming normal cases. It is further postulated that the AI's performance is comparable to human experts and that the AI “will likely complement rather than replace human experts” in veterinary radiology (1). This is a noteworthy finding given the scarcity of veterinary radiologists, and the prospect of the AI aiding or augmenting clinical practice. However, a closer examination reveals several methodological and interpretive limitations that temper these optimistic conclusions. This commentary discusses the most impactful concerns—from the study's lack of a true gold-standard reference and biased sample, to statistical and ethical issues—and highlights their implications for real-world veterinary radiology.

Background: Injuries of the sagittal groove of the proximal phalanx (P1) in equine athletes are considered to predominantly occur due to chronic bone stress overload.

Objectives: To describe the range of abnormalities that is present in the sagittal groove in a large group of horses diagnosed with sagittal groove disease (SGD) on low-field MRI.

Study Design: Retrospective, cross-sectional.

Methods: Medical records were searched to identify initial MRI images of horses diagnosed with SGD and these were blindly evaluated using a semi-quantitative grading scheme and novel SGD MRI classification system reflecting potential pathways of pathological progression and severity of stress injury.

Results: A total of 132 limbs from 111 horses were included in the study; predominantly warmbloods competing in showjumping (n = 83) and dressage (n = 18). SGD MRI classifications were: 0 (normal, n = 0), 1 (small subchondral defect, n = 2), 2 (osseous densification, n = 28), 3 (subchondral microfissure with osseous densification, n = 7), 4 (bone oedema-like signal within the subchondral +/- trabecular bone and +/- subchondral microfissure or demineralisation, n = 72), 5 (incomplete macrofissure/fracture, n = 23) and 6 (complete fracture, n = 0). Classification 4c (bone oedema-like signal with demineralisation) and 5 had higher proportions in the plantar third of hindlimbs (3% and 10%, respectively) compared with forelimbs (0% and 0%, respectively). SGD classification and extent of bone oedema-like signal were not significantly different between lame (n = 116) and non-lame limbs (n = 16) (both p > 0.05). Periosteal new bone and oedema-like signal were identified (either confidently or suspected) at the dorsoproximal aspect of P1 in 25% and 39% of limbs, respectively.

Main Limitations: Inclusion via diagnoses in original MRI reports, variable clinical history, small size of some classification groups.

Conclusions: The presence or absence of lameness is not a dependable measure of the severity of SGD. The periosteal oedema-like signal of P1 has not previously been described in MRI of SGD and further supports the concept of bone stress injury.

Background: Sagittal groove disease of the proximal phalanx in equine athletes is commonly considered a bone stress injury. Repetitive hyperextension of the fetlock under high load is thought to contribute to its development. Concurrent changes are often reported in the dorsal sagittal ridge of the third metacarpus/metatarsus (MC3/MT3).

Objectives: To describe the spectrum of associated osseous abnormalities that are present in the fetlock in a large group of horses diagnosed with sagittal groove disease on low-field magnetic resonance imaging (MRI).

Study design: Retrospective, cross-sectional.

Methods: MRI images of horses diagnosed with sagittal groove disease at Equitom Equine Clinic between March 2014 and March 2023 were evaluated using semi-quantitative grading schemes and a sagittal groove disease MRI classification system.

Results: MRIs of 132 limbs were evaluated, predominantly from warmbloods used for showjumping (n = 83) and dressage (n = 18). Osseous densification and bone oedema-like signal grades were higher in the dorsal sagittal ridge than palmarly/plantarly (p < 0.001 and p < 0.05, respectively). Grades of both osseous densification and bone oedema-like signal in the dorsal sagittal ridge did not significantly differ between the different sagittal groove disease MRI classifications (both p > 0.05).

Main limitations: Inclusion based on original MRI reports, absence of control group, small numbers within some grading groups hindering statistical analyses.

Conclusions: Findings support the aetiological theories of chronic bone-stress due to loaded fetlock hyperextension however the severity of osseous changes of the dorsal sagittal ridge does not appear to be associated with the severity of sagittal groove disease classification.

Evolution of magnetic resonance imaging (MRI) findings in horses with sagittal groove disease (SGD) of the proximal phalanx is relatively sparsely described. This retrospective, descriptive, longitudinal study describes the findings of sequential low-field MRI fetlock examinations in horses with SGD of the proximal phalanx using a classification system. Twenty-nine horses were included, predominantly warmbloods used for show jumping (79%). For 29 limbs re-examined during the initial rehabilitation period, classification remained constant (n = 18), increased (n = 2), decreased (n = 7), and fluctuated (n = 2). Notably, two limbs with initial classification 4b (bone oedema-like signal with subchondral microfissure) and one with 4c (bone oedema-like signal with subchondral demineralisation) progressed to classification 5 (incomplete macrofissure/fracture), highlighting their potential as prodromal or imminent fissure pathology. Following conservative (n = 28) and surgical (n = 1) treatment, 86% of the horses re-entered full training and competition with a mean +/- sd recovery time of 9.4 +/- 4.4 months. In total, 20% of horses in the study subsequently presented for repeat MRI due to recurrent lameness after resuming full work, with classification that was the same (n = 2), increased (n = 2), or decreased (n = 2) compared with the last scan. This study underscores the variability in progression of SGD MRI findings, emphasising the need for further larger-scale research into patterns of progression.

An 18-year-old female Fjord horse presented with a history of recent progressive dysphagia, coughing and increased respiratory effort during exercise, which had not improved with administration of systemic antimicrobials. Upper airway endoscopy, radiography and computed tomography showed a soft tissue mass at the dorsal aspect of the larynx in the region of the proximal cervical oesophagus. Humane euthanasia was opted for and post-mortem, histological and immunohistochemical examination provided a definitive diagnosis of leiomyoma of the proximal oesophagus. Neoplasia, and more specifically a smooth muscle neoplasm, of the proximal cervical oesophagus has not previously been reported in horses.

Metal artifacts in CT negatively impact the evaluation of surgical implants and the surrounding tissues. The aim of this prospective experimental study was to evaluate the ability of a single energy metal artifact reduction (SEMAR™, Canon) algorithm and virtual monoenergetic (VM) dual-energy CT (DECT) scanning techniques to reduce metal artifacts from stainless steel screws surgically inserted into the equine proximal phalanx. Seven acquisitions of 18 cadaver limbs were performed on a Canon Aquilion One Vision CT scanner (Helical +SEMAR, Volume +SEMAR, Standard Helical, Standard Volume and VM DECT at 135, 120, and 105 keV) and reconstructed in a bone kernel. Blinded subjective evaluation performed by three observers indicated a significant effect of acquisition in both adjacent tissues (P < 0.001) and distant tissues (P < 0.001) and the best metal artifact reduction was seen with Helical +SEMAR and Volume +SEMAR. The subjective overall preference of CT acquisition type was (1) Helical +SEMAR, (2) Volume +SEMAR, (3) VM DECT 135 keV, (4) VM DECT 120 keV, (5) VM DECT 105 keV, (6) Standard Helical, (7) Standard Volume (P < 0.001). Unblinded objective evaluation performed by one observer showed that VM DECT 120 keV, Helical +SEMAR, and Volume +SEMAR performed similarly and were objectively the best at reducing blooming artifact. Overall, the best metal artifact reduction was obtained with SEMAR, followed by VM DECT. However, VM DECT performance varies with energy level and was associated with decreased image quality in distant tissues and artifactual overcorrection of metal artifacts at high energy levels.

Background Forelimb extensor tendon rupture (ETR) in foals is sparsely documented.

Objectives To describe clinical and imaging findings, treatment details, and long-term outcome in foals with ETR.

Study design: Multicentre retrospective case-series.

Methods: Clinical record review of foals with ETR, presented to Faculty of Veterinary Medicine at Ghent University (Belgium), the Clinic for Horses at the University of Veterinary Medicine Hannover (Germany), and Rossdales Equine Hospital (UK) between 2009-2021. Long-term outcome based on structured telephone interview with the owner at the university hospitals and on future ratings from a public racing results archive at Rossdales Equine Hospital.

Results: Eight foals (28%; 95%CI: 11.7%-44.3%) presented with ETR only, whereas 21 foals (72%; 95%CI: 55.7%-88.3%) presented with ETR and concurrent flexural limb deformity foals with ETR only were older (median 20 days; Interquartile range [IQR] 13-22) than foals with ETR and flexural limb deformity (median five days; IQR 3-11; p = 0.001). Treatment included medical support, immobilization, and box rest. Hospitalisation time was longer for foals with ETR and flexural limb deformity (median 26 days; IQR 16-44) than for ETR-only foals (median 11 days; IQR 6-16; p = 0.03). Total rehabilitation time was longer for foals with ETR and flexural limb deformity (median 58 days; IQR 42-91) than for foals without flexural limb deformity (median 32 days; IQR 12-39; p = 0.03). Six foals (21%; 95%CI: 6.3%-35.7%) were euthanased within seven months: three because of problems with limb protraction and progressive flexural limb deformity. Long-term follow-up (>18 months) was available for two cases with ETR only, and seven cases with ETR and flexural limb deformity. Both ETR-only foals had a positive cosmetic outcome but only one a positive functional outcome. Foals with ETR and concurrent flexural limb deformity suffered limitations regarding cosmetic appearance (5/7 negative) and limb function (7/7 negative).

Main limitations: Small sample size, retrospective study, and no clinical follow-up.

Conclusions: ETR is a rare condition in this sample of foals treated under hospital conditions often requiring prolonged care, especially with concurrent flexural limb deformity. Further research should investigate potential associations with future athletic performance.

Background
Alpacas are being more frequently presented to veterinarians in the UK. It is important to validate whether published normal ocular parameters are consistent with the alpaca population in the UK.

Methods
Ophthalmic examinations were performed on healthy alpacas (Vicugna pacos) from three farms in East Anglia, UK.

Results
On direct ophthalmoscopy of 35 alpacas, there was a 50 per cent prevalence of opacities within the lens in alpacas older than two years old (n=8/16). There was a 36.8 per cent prevalence of persistent hyaloid arteries in alpacas under two years old (n=7/19). The mean Schirmer tear test‐1 value was 20.0 ±6 mm/minute (n=40). The mean intraocular pressure measured by rebound tonometry was 17.2 ±5.5 mmHg (n=46), and applanation tonometry resulted in statistically similar values (P=0.30; n=25). There was a significant variation in intraocular pressure throughout a 24‐hour period (n=8). Fluorescein dye was not detected at the nostrils of any of the alpacas which underwent a Jones test to assess nasolacrimal duct patency (n=8).

Conclusion
The ophthalmic findings appear largely consistent with previously published values from North America and continental Europe. Variations include the large range of measurements obtained and evidence of diurnal variation in intraocular pressure.