Department of Radiology, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden.
Department of Radiology, Region Västra Götaland, Sahlgrenska University Hospital, Gothenburg, Sweden.
Scand J Urol. 2021 Dec;55(6):427-433. doi: 10.1080/21681805.2021.1977845. Epub 2021 Sep 25.
Artificial intelligence (AI) offers new opportunities for objective quantitative measurements of imaging biomarkers from positron-emission tomography/computed tomography (PET/CT). Clinical image reporting relies predominantly on observer-dependent visual assessment and easily accessible measures like SUV, representing lesion uptake in a relatively small amount of tissue. Our hypothesis is that measurements of total volume and lesion uptake of the entire tumour would better reflect the disease`s activity with prognostic significance, compared with conventional measurements.
An AI-based algorithm was trained to automatically measure the prostate and its tumour content in PET/CT of 145 patients. The algorithm was then tested retrospectively on 285 high-risk patients, who were examined using F-choline PET/CT for primary staging between April 2008 and July 2015. Prostate tumour volume, tumour fraction of the prostate gland, lesion uptake of the entire tumour, and SUV were obtained automatically. Associations between these measurements, age, PSA, Gleason score and prostate cancer-specific survival were studied, using a Cox proportional-hazards regression model.
Twenty-three patients died of prostate cancer during follow-up (median survival 3.8 years). Total tumour volume of the prostate ( = 0.008), tumour fraction of the gland ( = 0.005), total lesion uptake of the prostate ( = 0.02), and age ( = 0.01) were significantly associated with disease-specific survival, whereas SUV ( = 0.2), PSA ( = 0.2), and Gleason score ( = 0.8) were not.
AI-based assessments of total tumour volume and lesion uptake were significantly associated with disease-specific survival in this patient cohort, whereas SUV and Gleason scores were not. The AI-based approach appears well-suited for clinically relevant patient stratification and monitoring of individual therapy.
人工智能（AI）为正电子发射断层扫描/计算机断层扫描（PET/CT）中的成像生物标志物的客观定量测量提供了新的机会。临床图像报告主要依赖于观察者依赖的视觉评估和易于获得的测量方法，如 SUV，代表相对少量组织中的病变摄取。我们的假设是，与传统测量相比，整个肿瘤的总体积和病变摄取的测量将更好地反映疾病的活性和预后意义。
我们训练了一种基于 AI 的算法，以自动测量 145 名患者的前列腺及其肿瘤含量。然后，该算法在 2008 年 4 月至 2015 年 7 月期间使用 F-胆碱 PET/CT 对 285 名高危患者进行的原发性分期检查中进行了回顾性测试。自动获得前列腺肿瘤体积、前列腺肿瘤分数、整个肿瘤的病变摄取和 SUV。使用 Cox 比例风险回归模型研究这些测量值与年龄、PSA、Gleason 评分和前列腺癌特异性生存之间的关系。
在随访期间，有 23 名患者死于前列腺癌（中位生存时间为 3.8 年）。前列腺总肿瘤体积（ = 0.008）、腺体肿瘤分数（ = 0.005）、前列腺总病变摄取（ = 0.02）和年龄（ = 0.01）与疾病特异性生存显著相关，而 SUV（ = 0.2）、PSA（ = 0.2）和 Gleason 评分（ = 0.8）则不相关。
在该患者队列中，基于 AI 的总肿瘤体积和病变摄取评估与疾病特异性生存显著相关，而 SUV 和 Gleason 评分则不相关。基于 AI 的方法似乎非常适合于临床相关的患者分层和个体治疗监测。
