Department of Radiology, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China.
Department of Radiology, The Affiliated Hospital of Nanjing University of Chinese Medicine, Nanjing, China.
Eur Radiol. 2023 Jul;33(7):5118-5130. doi: 10.1007/s00330-023-09433-2. Epub 2023 Feb 1.
To develop an artificial intelligence (AI) model for prostate segmentation and prostate cancer (PCa) detection, and explore the added value of AI-based computer-aided diagnosis (CAD) compared to conventional PI-RADS assessment.
A retrospective study was performed on multi-centers and included patients who underwent prostate biopsies and multiparametric MRI. A convolutional-neural-network-based AI model was trained and validated; the reliability of different CAD methods (concurrent read and AI-first read) were tested in an internal/external cohort. The diagnostic performance, consistency and efficiency of radiologists and AI-based CAD were compared.
The training/validation/internal test sets included 650 (400/100/150) cases from one center; the external test included 100 cases (25/25/50) from three centers. For diagnosis accuracy, AI-based CAD methods showed no significant differences and were equivalent to the radiologists in the internal test (127/150 vs. 130/150 vs. 125/150 for reader 1; 127/150 vs.132/150 vs. 131/150 for reader 2; all p > 0.05), whereas in the external test, concurrent-read methods were superior/equal to AI-first read (87/100 vs. 71/100, p < 0.001, for reader 2; 79/100 vs. 69/100, p = 0.076, for reader 1) and better than/equal to radiologists (79/100 vs. 72/100, p = 0.039, for reader 1; 87/100 vs. 86/100, p = 1.000, for reader 2). Moreover, AI-first read/concurrent read improved consistency in both internal test (κ = 1.000, 0.830) and external test (κ = 0.958, 0.713) compared to radiologists (κ = 0.747, 0.600); AI-first read method (8.54 s/7.66 s) was faster than readers (92.72 s/89.54 s) and concurrent-read method (29.15 s/28.92 s), respectively.
AI-based CAD could improve the consistency and efficiency for accurate diagnosis; the concurrent-read method could enhance the diagnostic capabilities of an inexperienced radiologist in unfamiliar situations.
• For prostate cancer segmentation, the performance of multi-small Vnet displays optimal compared to small Vnet and Vnet (DSC vs. DSC, p = 0.021; DSC vs. DSC, p < 0.001). • For prostate gland segmentation, the mean/median DSCs for fine and coarse segmentation were 0.91/0.91 and 0.88/0.89, respectively. Fine segmentation displays superior performance compared to coarse (DSC vs. DSC, p < 0.001). • For PCa diagnosis, AI-based CAD methods improve consistency in internal (κ = 1.000; 0.830) and external (κ = 0.958; 0.713) tests compared to radiologists (κ = 0.747; 0.600); the AI-first read (8.54 s/7.66 s) was faster than the readers (92.72 s/89.54 s) and the concurrent-read method (29.15 s/28.92 s).
开发一种用于前列腺分割和前列腺癌（PCa）检测的人工智能（AI）模型，并探索基于 AI 的计算机辅助诊断（CAD）与传统 PI-RADS 评估相比的附加价值。
对多中心进行回顾性研究，包括接受前列腺活检和多参数 MRI 的患者。训练和验证基于卷积神经网络的 AI 模型；在内部/外部队列中测试不同 CAD 方法（同时读取和 AI 首先读取）的可靠性。比较放射科医生和基于 AI 的 CAD 的诊断性能、一致性和效率。
训练/验证/内部测试集包括来自一个中心的 650 例（400/100/150）病例；外部测试集包括来自三个中心的 100 例（25/25/50）病例。在诊断准确性方面，基于 AI 的 CAD 方法与放射科医生无显著差异且等效，在内部测试中（读者 1：127/150 与 130/150 与 125/150；读者 2：127/150 与 132/150 与 131/150；均 p>0.05），而在外部测试中，同时读取方法优于 AI 首先读取（读者 2：87/100 与 71/100，p<0.001；读者 1：79/100 与 69/100，p=0.076）且优于放射科医生（读者 1：79/100 与 72/100，p=0.039；读者 2：87/100 与 86/100，p=1.000）。此外，AI 首先读取/同时读取与放射科医生相比，在内部测试（κ=1.000，0.830）和外部测试（κ=0.958，0.713）中均提高了一致性；AI 首先读取方法（8.54 s/7.66 s）比读者（92.72 s/89.54 s）和同时读取方法（29.15 s/28.92 s）更快。
基于 AI 的 CAD 可以提高准确诊断的一致性和效率；同时读取方法可以增强经验不足的放射科医生在不熟悉情况下的诊断能力。
• 对于前列腺癌分割，多小 Vnet 的性能优于小 Vnet 和 Vnet（DSC 与 DSC，p=0.021；DSC 与 DSC，p<0.001）。• 对于前列腺腺体分割，精细和粗分割的平均/中位数 DSCs 分别为 0.91/0.91 和 0.88/0.89，精细分割优于粗分割（DSC 与 DSC，p<0.001）。• 对于 PCa 诊断，与放射科医生相比，基于 AI 的 CAD 方法在内部（κ=1.000；0.830）和外部（κ=0.958；0.713）测试中提高了一致性；AI 首先读取（8.54 s/7.66 s）比读者（92.72 s/89.54 s）和同时读取方法（29.15 s/28.92 s）更快。
