Comparison of artificial intelligence to the veterinary radiologist's diagnosis of canine cardiogenic pulmonary edema

Abstract

Application of artificial intelligence (AI) to improve clinical diagnosis is a burgeoning field in human and veterinary medicine. The objective of this prospective, diagnostic accuracy study was to determine the accuracy, sensitivity, and specificity of an AI‐based software for diagnosing canine cardiogenic pulmonary edema from thoracic radiographs, using an American College of Veterinary Radiology‐certified veterinary radiologist's interpretation as the reference standard. Five hundred consecutive canine thoracic radiographs made after‐hours by a veterinary Emergency Department were retrieved. A total of 481 of 500 cases were technically analyzable. Based on the radiologist's assessment, 46 (10.4%) of these 481 dogs were diagnosed with cardiogenic pulmonary edema (CPE+). Of these cases, the AI software designated 42 of 46 as CPE+ and four of 46 as cardiogenic pulmonary edema negative (CPE−). Accuracy, sensitivity, and specificity of the AI‐based software compared to radiologist diagnosis were 92.3%, 91.3%, and 92.4%, respectively (positive predictive value, 56%; negative predictive value, 99%). Findings supported using AI software screening for thoracic radiographs of dogs with suspected cardiogenic pulmonary edema to assist with short‐term decision‐making when a radiologist is unavailable.

Artificial intelligence in veterinary diagnostic imaging: A literature review

“…Some of the currently available commercial AI systems for radiographic interpretation in veterinary radiology [71][72][73] are utilized in this more comprehensive fashion. Two of the reviewed papers 44,53 represent validations of these commercial platforms, though each focus on the detection of a single condition. While valuable and certainly steps in the right direction, these studies alone are not sufficient to understand the full breadth of capabilities of these AI systems, nor are they sufficient to compare performance between competitors.…”

“…AI reports were generated in 0.65 to 8.3 minutes with an average of 2.45 minutes. 44 Also noteworthy is the lack of inclusion of subsequent echocardiographic evaluation or clinical information (such as response to diuretic therapy) that could be considered to support cardiogenic origin of pulmonary edema. The authors conclude that AI is a suitable screening tool for cardiogenic pulmonary edema in canine populations with a low prevalence of disease.…”

“…Kim et al 44 . compared the performance of AI to veterinary radiologists in detecting canine cardiogenic pulmonary edema on thoracic radiographs using the interpretation of a board‐certified veterinary radiologist as the gold standard.…”

“…Three papers focus exclusively on the cardiac silhouette. [38][39][40] Four papers address pulmonary parenchymal changes: three using radiography and one using CT. [41][42][43][44] Five groups applied AI for comprehensive evaluation of canine and/or feline thoracic radiographs, including evaluation of pulmonary, cardiovascular, pleural, mediastinal and musculoskeletal structures, and one focused on the pleural space. [45][46][47][48][49] equivalent, and agreement between the two methods was 85%.…”

“…This corresponded to a 92.3% accuracy, 91.3% sensitivity, and 92.4% specificity. AI reports were generated in 0.65 to 8.3 minutes with an average of 2.45 minutes 44 . Also noteworthy is the lack of inclusion of subsequent echocardiographic evaluation or clinical information (such as response to diuretic therapy) that could be considered to support cardiogenic origin of pulmonary edema.…”