Sun Kai, Chen Yushi, Bai Bingqian, Gao Yanhua, Xiao Jiaying, Yu Gang
Department of Biomedical Engineering, School of Basic Medical Sciences, Central South University, Changsha 410013, China.
Department of Pathology, School of Basic Medical Sciences, Central South University, Changsha 410013, China.
Diagnostics (Basel). 2023 Mar 28;13(7):1277. doi: 10.3390/diagnostics13071277.
Current artificial intelligence (AI) in histopathology typically specializes on a single task, resulting in a heavy workload of collecting and labeling a sufficient number of images for each type of cancer. Heterogeneous transfer learning (HTL) is expected to alleviate the data bottlenecks and establish models with performance comparable to supervised learning (SL).
An accurate source domain model was trained using 28,634 colorectal patches. Additionally, 1000 sentinel lymph node patches and 1008 breast patches were used to train two target domain models. The feature distribution difference between sentinel lymph node metastasis or breast cancer and CRC was reduced by heterogeneous domain adaptation, and the maximum mean difference between subdomains was used for knowledge transfer to achieve accurate classification across multiple cancers.
HTL on 1000 sentinel lymph node patches (L-HTL-1000) outperforms SL on 1000 sentinel lymph node patches (L-SL-1-1000) (average area under the curve (AUC) and standard deviation of L-HTL-1000 vs. L-SL-1-1000: 0.949 ± 0.004 vs. 0.931 ± 0.008, value = 0.008). There is no significant difference between L-HTL-1000 and SL on 7104 patches (L-SL-2-7104) (0.949 ± 0.004 vs. 0.948 ± 0.008, value = 0.742). Similar results are observed for breast cancer. B-HTL-1008 vs. B-SL-1-1008: 0.962 ± 0.017 vs. 0.943 ± 0.018, value = 0.008; B-HTL-1008 vs. B-SL-2-5232: 0.962 ± 0.017 vs. 0.951 ± 0.023, value = 0.148.
HTL is capable of building accurate AI models for similar cancers using a small amount of data based on a large dataset for a certain type of cancer. HTL holds great promise for accelerating the development of AI in histopathology.
当前组织病理学中的人工智能（AI）通常专注于单一任务，这导致为每种癌症收集和标记足够数量图像的工作量很大。异构迁移学习（HTL）有望缓解数据瓶颈，并建立性能与监督学习（SL）相当的模型。
使用28,634个结肠直肠切片训练一个准确的源域模型。此外，使用1000个前哨淋巴结切片和1008个乳腺切片训练两个目标域模型。通过异构域自适应减少前哨淋巴结转移或乳腺癌与结直肠癌之间的特征分布差异，并使用子域之间的最大均值差异进行知识转移，以实现对多种癌症的准确分类。
在1000个前哨淋巴结切片上进行的HTL（L-HTL-1000）优于在1000个前哨淋巴结切片上进行的SL（L-SL-1-1000）（L-HTL-1000与L-SL-1-1000的曲线下平均面积（AUC）和标准差：0.949±0.004对0.931±0.008，p值 = 0.008）。在7104个切片上（L-SL-2-7104），L-HTL-1000与SL之间无显著差异（0.949±0.004对0.948±0.008，p值 = 0.742）。乳腺癌也观察到类似结果。B-HTL-1008与B-SL-1-1008：0.962±0.017对0.943±0.018，p值 = 0.008；B-HTL-1008与B-SL-2-5232：0.962±0.017对0.951±0.023，p值 = 0.148。
HTL能够基于某类癌症的大数据集，使用少量数据为相似癌症构建准确的AI模型。HTL在加速组织病理学中AI的发展方面具有巨大潜力。
