Constructing different machine learning models for identifying pelvic lipomatosis based on AI-assisted CT image feature recognition

Pelvic lipomatosis (PL) is a rare benign proliferative disease of adipose tissue, characterized by the excessive proliferation of non-encapsulated fat surrounding the rectum and bladder within the pelvic region [1]. Engels [2] first described this condition in 1959, and Fogg and Smyth [3] introduced the term pelvic lipomatosis in 1969 to define this pathology. The diagnosis of PL primarily relies on clinical symptoms and imaging findings. The symptoms of PL are mainly due to the compression of pelvic organs by the proliferative adipose tissue. Compression of the urinary system structures can lead to lower urinary tract irritative symptoms, such as dysuria, increased urinary frequency, and urgency, which may progress to obstructive hydronephrosis and potentially renal failure [4‐6]. Compression of the rectum may result in constipation, tenesmus, and narrowing of stools [7]. In a minority of patients, symptoms may include perineal pain and discomfort, and lower limb edema due to impaired venous return from vascular compression. Typical imaging findings include a pear-shaped bladder, elevation of the bladder and sigmoid colon, and narrowing of the rectosigmoid colon [8]. …