Şeker Mesut, Niazi M Khalid Khan, Chen Wei, Frankel Wendy L, Gurcan Metin N
Department of Electrical and Electronics Engineering, Dicle University, Diyarbakir 21280, Turkey.
Department of Pathology, The Ohio State University, Columbus, OH 43210, USA.
Cancers (Basel). 2025 Apr 7;17(7):1245. doi: 10.3390/cancers17071245.
BACKGROUND/OBJECTIVES: Identifying tumor budding (TB) in colorectal cancer (CRC) is vital for better prognostic assessment as it may signify the initial stage of metastasis. Despite its importance, TB detection remains challenging due to subjectivity in manual evaluations. Identifying TBs remains difficult, especially at high magnification levels, leading to inconsistencies in prognosis. To address these issues, we propose an automated system for TB classification using deep learning.
We trained a deep learning model to identify TBs through weakly supervised learning by aggregating positive and negative bags from the tumor invasive front. We assessed various foundation models for feature extraction and compared their performance. Attention heatmaps generated by attention-based multi-instance learning (ABMIL) were analyzed to verify alignment with TBs, providing insights into the interpretability of the features. The dataset includes 29 WSIs for training and 70 whole slide images (WSIs) for the hold-out test set.
In six-fold cross-validation, Phikon-v2 achieved the highest average AUC (0.984 ± 0.003), precision (0.876 ± 0.004), and recall (0.947 ± 0.009). Phikon-v2 again achieved the highest AUC (0.979) and precision (0.980) on the external hold-out test set. Moreover, its recall rate (0.910) was still higher than that of UNI's (0.879). UNI exhibited a balanced performance on the hold-out test set, with an AUC of 0.960 and a precision of 0.968. CtransPath showed strong precision on the external hold-out test set (0.947) but had a slightly lower recall (0.911).
The proposed technique enhances the accuracy of TB assessment, offering potential applications for CRC and other cancer types.
背景/目的：识别结直肠癌（CRC）中的肿瘤芽生（TB）对于更好的预后评估至关重要，因为它可能标志着转移的初始阶段。尽管其很重要，但由于手动评估的主观性，TB检测仍然具有挑战性。识别TB仍然很困难，尤其是在高放大倍数水平下，这导致预后的不一致性。为了解决这些问题，我们提出了一种使用深度学习的TB分类自动化系统。
我们通过聚合肿瘤浸润前沿的正样本和负样本袋，训练了一个深度学习模型，通过弱监督学习来识别TB。我们评估了各种用于特征提取的基础模型，并比较了它们的性能。分析了基于注意力的多实例学习（ABMIL）生成的注意力热图，以验证与TB的对齐情况，从而深入了解特征的可解释性。该数据集包括29张用于训练的全切片图像（WSI）和70张用于保留测试集的全切片图像（WSI）。
在六折交叉验证中，Phikon-v2实现了最高的平均AUC（0.984±0.003）、精度（0.876±0.004）和召回率（0.947±0.009）。在外部保留测试集上，Phikon-v2再次实现了最高的AUC（0.979）和精度（0.980）。此外，其召回率（0.910）仍然高于UNI的召回率（0.879）。UNI在保留测试集上表现出平衡的性能，AUC为0.960，精度为0.968。CtransPath在外部保留测试集上表现出很强的精度（0.947），但召回率略低（0.911）。
所提出的技术提高了TB评估的准确性，为CRC和其他癌症类型提供了潜在的应用。
