Detecting Uteropelvic Junction Obstruction at an Early Stage using Ultrasound Images through an Interplanetary File System

Ureteropelvic junction obstruction (UPJO) is a medical condition characterized by a blockage at the junction between the ureter and the kidney, leading to impaired urine flow and increased fluid pressure. If left untreated, UPJO can progress to hydronephrosis and potentially result in irreversible kidney damage. This paper presents a novel system for the early detection and severity grading of UPJO using Deep Learning Computer-Aided Diagnosis (DL-CAD) techniques. The proposed approach leverages ultrasound imaging as a primary diagnostic tool, noted for its cost-effectiveness, absence of radiation, and non-invasive nature. To enhance the security and integrity of medical data transmission, we integrate blockchain technology into our framework. Our solution combines a DL- CAD model featuring an attention-based pyramid semantic segmentation network with a discrete wavelet transformation- processed residual classification network. An IoT device is employed to detect patient abnormalities and transmit this data to the diagnosis model via an InterPlanetary File System (IPFS). This decentralized storage solution facilitates secure data management across a peer-to-peer network. Ultimately, the proposed framework not only identifies the presence of UPJO but also assesses its severity, ensuring that patient data remains protected throughout the diagnostic process.