Unfiltered Sonographic Images for Kidney Study(USIK)

A.Used Database:This study was conducted using kidney ultrasound images acquired from the Radiology Department of Kosar Hospital, affiliated with Semnan University of Medical Sciences in Semnan, Iran. A total of 47 patients (11 males and 36 females), aged between 30 and 70 years and weighing between 60 and 100 kg, were examined Ultrasound imaging was performed using a Vinno G8 system (China) equipped with a 3.5 MHz convex probe, calibrated monthly to ensure imaging accuracy. All scans were conducted by a single operator with over 10 years of clinical experience, under controlled room conditions (24 ± 2°C), following written informed consent from participants. None of the patients had a history of kidney surgery, transplantation, or congenital anomalies such as malrotation, horseshoe kidney, or ectopic kidney. To optimize imaging, patients performed deep breathing exercises before positioning. The right kidney was scanned in a supine position, while the left kidney was examined in a left semi-lateral position. Each scan lasted approximately 10 minutes, capturing 6–8 frames per kidney under optimal abdominal conditions to minimize interference from bowel gas. Ultrasound settings were standardized across all cases: a 15 cm imaging depth, no zoom, and body habitus-adjusted gain. The imaging mode employed CEVINNO HAR (harmonic imaging) with a 4.5 MHz transducer frequency, AP100% power, DG46% gain, and a 25.3 Hz frame rate; this mode is particularly effective for improving stone identification by reducing artifacts and enhancing tissue contrast. The S1-8C curved array probe and ABD preset were used, ensuring appropriate visualization of renal structures. Harmonic imaging in this convex probe is particularly effective for identifying kidney stones by reducing artifacts and enhancing tissue contrast. Safety indices remained within recommended limits (MI: 1.4, TIS: 0.7), adhering to clinical guidelines. These parameters supported high-resolution imaging of the renal parenchyma, collecting systems, and surrounding anatomy. The average kidney length was 10.5 ± 1.2 cm, with a parenchymal thickness of 1.5 ± 0.3 cm. Images with severe gas artifacts were The main clinical indications for scanning included physician-requested evaluation of renal function, such as flank pain, urinary problems, periodic check-ups, and other related concerns. Image acquisition followed a sweeping motion along the kidney’s mid-sagittal plane until both poles were visualized. Ultrasound beams were oriented perpendicularly to the kidney’s long axis to enhance corticomedullary border visibility. No significant hydronephrosis was observed among the cases, except for one or two instances of mild dilation of the pelvicalyceal system. All imaging data were archived for use in developing artificial intelligence (AI) models. The strict protocol ensured consistent, high-quality images suitable for both clinical evaluation and research applications. Patients with a BMI over 35 were excluded to maintain imaging quality, which is noted as a primary limitation of the dataset. By adhering to standardized techniques and excluding anatomical abnormalities, the study generated a robust dataset optimized for AI training and diagnostic reliability.