Ruggiero Santeramo, Celeste Damiani, Jiefei Wei, Giovanni Montana, Adam R Brentnall
{"title":"更好的乳腺癌检测人工智能算法是否也能更好地预测风险?一项配对病例对照研究。","authors":"Ruggiero Santeramo, Celeste Damiani, Jiefei Wei, Giovanni Montana, Adam R Brentnall","doi":"10.1186/s13058-024-01775-z","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>There is increasing evidence that artificial intelligence (AI) breast cancer risk evaluation tools using digital mammograms are highly informative for 1-6 years following a negative screening examination. We hypothesized that algorithms that have previously been shown to work well for cancer detection will also work well for risk assessment and that performance of algorithms for detection and risk assessment is correlated.</p><p><strong>Methods: </strong>To evaluate our hypothesis, we designed a case-control study using paired mammograms at diagnosis and at the previous screening visit. The study included n = 3386 women from the OPTIMAM registry, that includes mammograms from women diagnosed with breast cancer in the English breast screening program 2010-2019. Cases were diagnosed with invasive breast cancer or ductal carcinoma in situ at screening and were selected if they had a mammogram available at the screening examination that led to detection, and a paired mammogram at their previous screening visit 3y prior to detection when no cancer was detected. Controls without cancer were matched 1:1 to cases based on age (year), screening site, and mammography machine type. Risk assessment was conducted using a deep-learning model designed for breast cancer risk assessment (Mirai), and three open-source deep-learning algorithms designed for breast cancer detection. Discrimination was assessed using a matched area under the curve (AUC) statistic.</p><p><strong>Results: </strong>Overall performance using the paired mammograms followed the same order by algorithm for risk assessment (AUC range 0.59-0.67) and detection (AUC 0.81-0.89), with Mirai performing best for both. There was also a correlation in performance for risk and detection within algorithms by cancer size, with much greater accuracy for large cancers (30 mm+, detection AUC: 0.88-0.92; risk AUC: 0.64-0.74) than smaller cancers (0 to < 10 mm, detection AUC: 0.73-0.86, risk AUC: 0.54-0.64). Mirai was relatively strong for risk assessment of smaller cancers (0 to < 10 mm, risk, Mirai AUC: 0.64 (95% CI 0.57 to 0.70); other algorithms AUC 0.54-0.56).</p><p><strong>Conclusions: </strong>Improvements in risk assessment could stem from enhancing cancer detection capabilities of smaller cancers. Other state-of-the-art AI detection algorithms with high performance for smaller cancers might achieve relatively high performance for risk assessment.</p>","PeriodicalId":49227,"journal":{"name":"Breast Cancer Research","volume":null,"pages":null},"PeriodicalIF":7.4000,"publicationDate":"2024-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10848404/pdf/","citationCount":"0","resultStr":"{\"title\":\"Are better AI algorithms for breast cancer detection also better at predicting risk? A paired case-control study.\",\"authors\":\"Ruggiero Santeramo, Celeste Damiani, Jiefei Wei, Giovanni Montana, Adam R Brentnall\",\"doi\":\"10.1186/s13058-024-01775-z\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Background: </strong>There is increasing evidence that artificial intelligence (AI) breast cancer risk evaluation tools using digital mammograms are highly informative for 1-6 years following a negative screening examination. We hypothesized that algorithms that have previously been shown to work well for cancer detection will also work well for risk assessment and that performance of algorithms for detection and risk assessment is correlated.</p><p><strong>Methods: </strong>To evaluate our hypothesis, we designed a case-control study using paired mammograms at diagnosis and at the previous screening visit. The study included n = 3386 women from the OPTIMAM registry, that includes mammograms from women diagnosed with breast cancer in the English breast screening program 2010-2019. Cases were diagnosed with invasive breast cancer or ductal carcinoma in situ at screening and were selected if they had a mammogram available at the screening examination that led to detection, and a paired mammogram at their previous screening visit 3y prior to detection when no cancer was detected. Controls without cancer were matched 1:1 to cases based on age (year), screening site, and mammography machine type. Risk assessment was conducted using a deep-learning model designed for breast cancer risk assessment (Mirai), and three open-source deep-learning algorithms designed for breast cancer detection. Discrimination was assessed using a matched area under the curve (AUC) statistic.</p><p><strong>Results: </strong>Overall performance using the paired mammograms followed the same order by algorithm for risk assessment (AUC range 0.59-0.67) and detection (AUC 0.81-0.89), with Mirai performing best for both. There was also a correlation in performance for risk and detection within algorithms by cancer size, with much greater accuracy for large cancers (30 mm+, detection AUC: 0.88-0.92; risk AUC: 0.64-0.74) than smaller cancers (0 to < 10 mm, detection AUC: 0.73-0.86, risk AUC: 0.54-0.64). Mirai was relatively strong for risk assessment of smaller cancers (0 to < 10 mm, risk, Mirai AUC: 0.64 (95% CI 0.57 to 0.70); other algorithms AUC 0.54-0.56).</p><p><strong>Conclusions: </strong>Improvements in risk assessment could stem from enhancing cancer detection capabilities of smaller cancers. Other state-of-the-art AI detection algorithms with high performance for smaller cancers might achieve relatively high performance for risk assessment.</p>\",\"PeriodicalId\":49227,\"journal\":{\"name\":\"Breast Cancer Research\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":7.4000,\"publicationDate\":\"2024-02-07\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10848404/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Breast Cancer Research\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1186/s13058-024-01775-z\",\"RegionNum\":1,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"Medicine\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Breast Cancer Research","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1186/s13058-024-01775-z","RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"Medicine","Score":null,"Total":0}
引用次数: 0
摘要
背景:越来越多的证据表明,人工智能(AI)乳腺癌风险评估工具使用数字乳房X光照片在筛查阴性后的1-6年内具有很高的信息量。我们假设,以前在癌症检测方面表现良好的算法在风险评估方面也会表现良好,而且检测算法和风险评估算法的表现是相关的:为了评估我们的假设,我们设计了一项病例对照研究,使用诊断时和上次筛查时的成对乳房 X 线照片。研究对象包括来自 OPTIMAM 登记处的 n = 3386 名妇女,其中包括在 2010-2019 年英国乳腺癌筛查计划中被诊断为乳腺癌的妇女的乳房 X 光照片。病例在筛查时被确诊为浸润性乳腺癌或导管原位癌,如果她们在筛查时有乳房X光照片导致检测,并且在检测前3年的上一次筛查中没有检测到癌症,则被选中。未患癌症的对照组与病例根据年龄(年)、筛查地点和乳腺 X 射线摄影机类型进行 1:1 匹配。使用专为乳腺癌风险评估设计的深度学习模型(Mirai)和三种专为乳腺癌检测设计的开源深度学习算法进行风险评估。使用匹配的曲线下面积(AUC)统计来评估识别率:结果:使用配对乳房X光照片的总体性能在风险评估(AUC范围为0.59-0.67)和检测(AUC为0.81-0.89)方面遵循相同的算法顺序,Mirai在这两个方面的性能最好。不同癌症大小的算法在风险和检测方面的表现也存在相关性,大癌(30 毫米以上,检测 AUC:0.88-0.92;风险 AUC:0.64-0.74)的准确性远高于小癌(0 至 < 10 毫米,检测 AUC:0.73-0.86;风险 AUC:0.54-0.64)。Mirai 对较小癌症的风险评估能力相对较强(0 至 < 10 毫米,风险,Mirai AUC:结论:结论:风险评估的改进可能源于提高对较小癌症的检测能力。其他最先进的人工智能检测算法对较小的癌症具有较高的检测性能,在风险评估中也可能达到相对较高的性能。
Are better AI algorithms for breast cancer detection also better at predicting risk? A paired case-control study.
Background: There is increasing evidence that artificial intelligence (AI) breast cancer risk evaluation tools using digital mammograms are highly informative for 1-6 years following a negative screening examination. We hypothesized that algorithms that have previously been shown to work well for cancer detection will also work well for risk assessment and that performance of algorithms for detection and risk assessment is correlated.
Methods: To evaluate our hypothesis, we designed a case-control study using paired mammograms at diagnosis and at the previous screening visit. The study included n = 3386 women from the OPTIMAM registry, that includes mammograms from women diagnosed with breast cancer in the English breast screening program 2010-2019. Cases were diagnosed with invasive breast cancer or ductal carcinoma in situ at screening and were selected if they had a mammogram available at the screening examination that led to detection, and a paired mammogram at their previous screening visit 3y prior to detection when no cancer was detected. Controls without cancer were matched 1:1 to cases based on age (year), screening site, and mammography machine type. Risk assessment was conducted using a deep-learning model designed for breast cancer risk assessment (Mirai), and three open-source deep-learning algorithms designed for breast cancer detection. Discrimination was assessed using a matched area under the curve (AUC) statistic.
Results: Overall performance using the paired mammograms followed the same order by algorithm for risk assessment (AUC range 0.59-0.67) and detection (AUC 0.81-0.89), with Mirai performing best for both. There was also a correlation in performance for risk and detection within algorithms by cancer size, with much greater accuracy for large cancers (30 mm+, detection AUC: 0.88-0.92; risk AUC: 0.64-0.74) than smaller cancers (0 to < 10 mm, detection AUC: 0.73-0.86, risk AUC: 0.54-0.64). Mirai was relatively strong for risk assessment of smaller cancers (0 to < 10 mm, risk, Mirai AUC: 0.64 (95% CI 0.57 to 0.70); other algorithms AUC 0.54-0.56).
Conclusions: Improvements in risk assessment could stem from enhancing cancer detection capabilities of smaller cancers. Other state-of-the-art AI detection algorithms with high performance for smaller cancers might achieve relatively high performance for risk assessment.
期刊介绍:
Breast Cancer Research, an international, peer-reviewed online journal, publishes original research, reviews, editorials, and reports. It features open-access research articles of exceptional interest across all areas of biology and medicine relevant to breast cancer. This includes normal mammary gland biology, with a special emphasis on the genetic, biochemical, and cellular basis of breast cancer. In addition to basic research, the journal covers preclinical, translational, and clinical studies with a biological basis, including Phase I and Phase II trials.