The pathology is decisive for disease diagnosis but relies heavily on experienced pathologists. In recent years, there has been growing interest in the use of artificial intelligence in pathology (AIP) to enhance diagnostic accuracy and efficiency. However, the impressive performance of deep learning-based AIP in laboratory settings often proves challenging to replicate in clinical practice. As the data preparation is important for AIP, the paper has reviewed AIP-related studies in the PubMed database published from January 2017 to February 2022, and 118 studies were included. An in-depth analysis of data preparation methods is conducted, encompassing the acquisition of pathological tissue slides, data cleaning, screening, and subsequent digitization. Expert review, image annotation, dataset division for model training and validation are also discussed. Furthermore, we delve into the reasons behind the challenges in reproducing the high performance of AIP in clinical settings and present effective strategies to enhance AIP’s clinical performance. The robustness of AIP depends on a randomized collection of representative disease slides, incorporating rigorous quality control and screening, correction of digital discrepancies, reasonable annotation, and sufficient data volume. Digital pathology is fundamental in clinical-grade AIP, and the techniques of data standardization and weakly supervised learning methods based on whole slide image (WSI) are effective ways to overcome obstacles of performance reproduction. The key to performance reproducibility lies in having representative data, an adequate amount of labeling, and ensuring consistency across multiple centers. Digital pathology for clinical diagnosis, data standardization and the technique of WSI-based weakly supervised learning will hopefully build clinical-grade AIP.

Background: 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). Methods: 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. Result: 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, p 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, p 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, p value = 0.008; B-HTL-1008 vs. B-SL-2-5232: 0.962 ± 0.017 vs. 0.951 ± 0.023, p value = 0.148. Conclusions: 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.

Objective: In Transthoracic echocardiographic (TTE) examination, it is essential to identify the cardiac views accurately. Computer-aided recognition is expected to improve the accuracy of the TTE examination. Methods: This paper proposes a new method for automatic recognition of cardiac views based on deep learning, including three strategies. First, A spatial transform network is performed to learn cardiac shape changes during the cardiac cycle, which reduces intra-class variability. Second, a channel attention mechanism is introduced to adaptively recalibrates channel-wise feature responses. Finally, unlike conventional deep learning methods, which learned each input images individually, the structured signals are applied by a graph of similarities among images. These signals are transformed into the graph-based image embedding, which act as unsupervised regularization constraints to improve the generalization accuracy. Results: The proposed method was trained and tested in 171792 cardiac images from 584 subjects. Compared with the known result of the state of the art, the overall accuracy of the proposed method on cardiac image classification is 99.10% vs. 91.7%, and the mean AUC is 99.36%. Moreover, the overall accuracy is 98.15%, and the mean AUC is 98.96% on an independent test set with 34211 images from 100 subjects. Conclusion: The method of this paper achieved the results of the state of the art, which is expected to be an automated recognition tool for cardiac views recognition. The work confirms the potential of deep learning on ultrasound medicine.

Background: The successful application of deep learning in medical images requires a large amount of annotation data for supervised training. However, massive labeling of medical data is expensive and time consuming. This paper proposes a semi-supervised deep learning method for the detection and classification of benign and malignant breast nodules in ultrasound images, which include two phases. Methods: The nodule position in the ultrasound image is firstly detected using the faster RCNN network. Second, the recognition network is used to identify the benign and malignant types of nodules. The method in this paper uses a semi-supervised learning strategy, using 800 labeled nodules and 4396 unlabeled nodules. Results: Based on mean teacher training strategy, the proposed semi-supervised network has obtained excellent results, which is similar to currently used with supervised training networks. On the two test data sets, the AUC of semi-supervised learning and supervised learning were: 93.7% vs 94.2% and 92% vs 92.3%. Conclusions: The paper proves that semi-supervised learning strategies have good application potential in medical images. Based on a special learning strategy, the result of semi-supervised learning is expected to achieve close or even achieve similar result of supervised deep learning, which only need a small number of labeled samples and a large number of unlabeled samples. It means deep learning analysis of breast lesion will be more feasible and more efficient.