ABSTRACT
Objective:To summarize our preliminary experience of the individual transurethral en bloc resection of bladder tumor (ERBT) based on vesical imaging-reporting and data system (VI-RADS).Methods:The clinical data of 32 bladder cancer patients admitted from January 2019 to October 2019 were retrospectively analyzed, including 26 males and 6 females. Among them, there were 27, 5, 26 and 6 patients who had primary, recurrent, single or mutiple blader tumors, respectively. And the median number of bladder tumor was 1(1-3) and the mean diameter was 2(0.6-4.5)cm.The patients were aged 37 to 82 years, with a median age of 63 years. All patients underwent multi-parameter magnetic resonance imaging (mpMRI) before surgery and acquired a VI-RADS score. Among the 32 patients, there were 8, 17, 2, 5, and 0 patients in the VI-RADS score category 1, 2, 3, 4, and 5, respectively. Based on the VI-RADS score and tumor size, morphology and number provided by the mpMRI, the urologists classified the tumor types into type 1, 2a, 2b, 2c, 3a, 3b, 3c, 4a, 4b or 5, and designed the surgical protocol for each type including the resection plan, boundary and depth. There were 8, 6, 7, 4, 0, 1, 1, 3, 2 and 0 patients in each type, respectively. The tumor types were further confirmed during the operation, and the operation was completed according to the surgical plans for different tumor types.Patients received intravesical therapy of gemcitabine within 24 hours after surgery.Results:All operations were successfully completed and none was converted to the traditional transurethral resection of the bladder tumor. The operation time was 5 to 35 minutes with a median time of 15 minutes. Tumor specimens from all patients contained the muscularis propria. Among the patients with scores 1, 2, 3 and 4, there were 8, 16, 1 and 0 patients diagnosed with non-muscle invasive bladder cancer (NMIBC), respectively. All the patients with NMIBC had negative basal resection margins and 6 out 7 muscle invasive bladder cancer (MIBC) patients had negative resection margins. There were no intraoperative complications such as bladder perforation and obturator reflex. Four patients experienced obvious postoperative bladder irritation and relieved after symptomatic treatment or removing catheter. Twelve patients received second resections, including 10 NMIBC patients and 2 MIBC patients. No residual tumor was found in the re-resected specimens. There were 9 and 12 NMIBC patients received regular intravesical therapy of gemcitabine or BCG, respectively. Among the 7 MIBC patients, 5 received radical cystectomy and two received bladder-preserving treatment including second resection, adjuvant chemotherapy and radiotherapy. The follow-up period was 3-12 months, with a median of 6 months. One NMIBC patient relapsed at 9th months after surgery and underwent ERBT.Conclusions:The personalized ERBT based on VI-RADS is safe and feasible, and can achieve negative margins in all NMIBC and some MIBC without severe complications.
ABSTRACT
Artificial intelligence can potentially provide a substantial role in streamlining chest computed tomography (CT) diagnosis of COVID-19 patients. However, several critical hurdles have impeded the development of robust AI model, which include deficiency, isolation, and heterogeneity of CT data generated from diverse institutions. These bring about lack of generalization of AI model and therefore prevent it from applications in clinical practices. To overcome this, we proposed a federated learning-based Unified CT-COVID AI Diagnostic Initiative (UCADI, http://www.ai-ct-covid.team/), a decentralized architecture where the AI model is distributed to and executed at each host institution with the data sources or client ends for training and inferencing without sharing individual patient data. Specifically, we firstly developed an initial AI CT model based on data collected from three Tongji hospitals in Wuhan. After model evaluation, we found that the initial model can identify COVID from Tongji CT test data at near radiologist-level (97.5% sensitivity) but performed worse when it was tested on COVID cases from Wuhan Union Hospital (72% sensitivity), indicating a lack of model generalization. Next, we used the publicly available UCADI framework to build a federated model which integrated COVID CT cases from the Tongji hospitals and Wuhan Union hospital (WU) without transferring the WU data. The federated model not only performed similarly on Tongji test data but improved the detection sensitivity (98%) on WU test cases. The UCADI framework will allow participants worldwide to use and contribute to the model, to deliver a real-world, globally built and validated clinic CT-COVID AI tool. This effort directly supports the United Nations Sustainable Development Goals number 3, Good Health and Well-Being, and allows sharing and transferring of knowledge to fight this devastating disease around the world.
