1.Application of holographic image in transperineal prostate targeted biopsy
Lei WANG ; Zichen ZHAO ; Hongfeng GUO ; Manli NA ; Mengshen LI ; Yi WANG ; Ningchen LI ; Yanqun NA
Chinese Journal of Urology 2022;43(2):111-115
Objective:To investigate the feasibility and accuracy of transperineal prostate targeted biopsy guided by holographic image.Methods:Clinical data of 10 patients with transperineal prostate targeted biopsy guided by holographic image in Peking University Shougang Hospital between May and September 2020 were analyzed retrospectively. The average age was (70.9±10.3) years old, the median PSA was 15.1(6.02-1110.14) ng/ml, prostate MRI were performed before biopsy and the PI-RADS scores were all ≥ 3, and the number of suspicious target lesions was 1.4±0.5. CT examinations of urinary system were performed on the premise of mild lithotomy position and positioning stickers pasted on the skin of perineum and lower abdomen. The original data of CT and MRI were obtained, holographic image models were firstly made separately and then fused into a complete model, and the puncture paths were planned for the target lesions. At the time of puncture, the patient took the same body position as in CT scan, the operator wore a mixed reality head mounted display (HoloLens glasses), and the skin positioning stickers were used for visual registration between the holographic model and the real human body. Then under the guidance of the virtual puncture path, the puncture biopsy gun was placed, fired after reaching the predetermined depth, a transrectal ultrasound probe was placed to clarify the position of the puncture needle, and the objective accuracy of puncture was judged by comparison of ultrasound and MRI images. If the first shot was judged to be inaccurate, it was allowed to make a supplementary shot after adjusting the angle. After holographic guided biopsies, cognitive fusion targeted biopsies and 12-needle systematic biopsies were performed routinely, and the proportion of positive needles of the three different biopsy methods were calculated respectively.Results:All the 10 cases were successfully completed, including 16 holographic image guided shots, 28 cognitive fusion targeted shots and 116 systematic shots. The objective accuracy of holographic image guided biopsy after first shot judgments was 68.8% (11/16), while it raised to 87.5% (14/16) after supplementary shots. The proportion of positive needles in the three puncture methods were 56.3% (9/16), 42.9% (12/28) and 19.8% (23/116), respectively ( P=0.002). The results of subjective questionnaire showed that holographic model was helpful to improve the spatial understanding of lesions. The satisfaction of intraoperative holographic registration and guided puncture were 90% and 60%, respectively. No puncture related complication occurred in this group. Conclusion:The study preliminarily confirmed the feasibility of holographic image-guided prostate targeted biopsy. This new puncture method has better objective accuracy, and the proportion of positive needles is significantly better than systematic biopsy.
2.Updated Bayesian Network MetaAnalysis of Adjuvant Targeted Treatment Regimens for Early Human Epidermal Growth Factor Receptor-2 Positive Breast Cancer
Xinyan LI ; Litong YAO ; Mozhi WANG ; Mengshen WANG ; Xiang LI ; Xueting YU ; Jingyi GUO ; Haoran DONG ; Xiangyu SUN ; Yingying XU
Journal of Breast Cancer 2020;23(4):410-429
Purpose:
Combining targeted agents with adjuvant chemotherapy prolongs survival in human epidermal growth factor receptor 2 (HER2)-positive breast cancer patients, but also increases the risk of adverse effects. The updated results of 3 randomized controlled trials (RCTs) were reported in 2019. Given the lack of adequate head-to-head pairwise assessment for anti-HER2 agents, network meta-analysis facilitates obtaining more precise inference for evidence-based therapy.
Methods:
RCTs comparing at least 2 anti-HER2 regimens in an adjuvant setting for HER2-positive early-stage breast cancer (EBC) were included. Hazard ratios for overall survival (OS) and disease free survival (DFS), with respective 95% confidence intervals were pooled for assessment of efficacy. A Bayesian statistical model was used, and odds ratios (ORs) for adverse events (AEs) were used to pool effect sizes.
Results:
We demonstrated that 1-year trastuzumab plus chemotherapy had increased efficacy compared to shorter or longer treatment duration. The OR of cardiac events gradually increased from 6 months to 1 and 2-year trastuzumab arms, relative to chemotherapy only.Compared to trastuzumab plus chemotherapy, dual HER2-targeting therapies increased DFS, especially for hormone receptor negative patients. Dual anti-HER2 blockade regimens revealed an increased probability of gastrointestinal reactions. As a second agent, pertuzumab showed significantly higher DFS and OS.
Conclusion
We conclude that 1-year adjuvant trastuzumab should remain as the standard treatment for HER2-positive EBC patients, as it has greater efficacy and a manageable proportion of AEs. Clinical efficacy can be increased for hormone receptor-negative tumors by including a second HER2-targeted agent to the treatment regimen. For hormone receptorpositive cases with basal disease, it is acceptable to reduce the risk of cardiotoxicity by shortening the duration of trastuzumab.
4.Survey on natural language processing in medical image analysis.
Zhengliang LIU ; Mengshen HE ; Zuowei JIANG ; Zihao WU ; Haixing DAI ; Lian ZHANG ; Siyi LUO ; Tianle HAN ; Xiang LI ; Xi JIANG ; Dajiang ZHU ; Xiaoyan CAI ; Bao GE ; Wei LIU ; Jun LIU ; Dinggang SHEN ; Tianming LIU
Journal of Central South University(Medical Sciences) 2022;47(8):981-993
Recent advancement in natural language processing (NLP) and medical imaging empowers the wide applicability of deep learning models. These developments have increased not only data understanding, but also knowledge of state-of-the-art architectures and their real-world potentials. Medical imaging researchers have recognized the limitations of only targeting images, as well as the importance of integrating multimodal inputs into medical image analysis. The lack of comprehensive surveys of the current literature, however, impedes the progress of this domain. Existing research perspectives, as well as the architectures, tasks, datasets, and performance measures examined in the present literature, are reviewed in this work, and we also provide a brief description of possible future directions in the field, aiming to provide researchers and healthcare professionals with a detailed summary of existing academic research and to provide rational insights to facilitate future research.
Humans
;
Natural Language Processing
;
Surveys and Questionnaires