Artificial intelligence (AI), particularly deep learning algorithms, is gaining extensive attention for its excellent performance in image-recognition tasks. They can automatically make a quantitative assessment of complex medical image characteristics and achieve an increased accuracy for diagnosis with higher efficiency. AI is widely used and getting increasingly popular in the medical imaging of the liver, including radiology, ultrasound, and nuclear medicine. AI can assist physicians to make more accurate and reproductive imaging diagnosis and also reduce the physicians’ workload. This article illustrates basic technical knowledge about AI, including traditional machine learning and deep learning algorithms, especially convolutional neural networks, and their clinical application in the medical imaging of liver diseases, such as detecting and evaluating focal liver lesions, facilitating treatment, and predicting liver treatment response. We conclude that machine-assisted medical services will be a promising solution for future liver medical care. Lastly, we discuss the challenges and future directions of clinical application of deep learning techniques.

BackgroundEpithelial–mesenchymal transition (EMT) is implicated in the metastatic process and presents a challenge to epithelial cell adhesion molecule-based detection of circulating tumor cells (CTCs), which have been demonstrated to be a prognostic indicator in metastatic breast cancer. Although evidence has indicated that heterogeneity of CTCs based on EMT markers is associated with disease progression, no standard recommendations have been established for clinical practice. This study aimed to evaluate the prognostic significance of dynamic CTC detection based on EMT for metastatic breast cancer patients.MethodsWe enrolled 108 human epidermal growth factor receptor 2-negative metastatic breast cancer patients from the prospective phase III CAMELLIA study and applied the CanPatrol CTC enrichment technique to identify CTC phenotypes (including epithelial CTCs, biphenotypic epithelial/mesenchymal CTCs, and mesenchymal CTCs) in peripheral blood samples. Receiver operating characteristic curve analyses of total CTC count and the proportion of mesenchymal CTCs for predicting the 1-year progression-free survival (PFS) rate were conducted to determine the optimal cut-off values, and Kaplan–Meier analysis and Cox proportional hazards regression analysis were performed to investigate the prognostic value of the cut-off values of both total CTC count and the proportion of mesenchymal CTCs in combination.ResultsFor predicting the 1-year PFS rate, the optimal cut-off value of total CTC count was 9.5 (Area under the curve [AUC] = 0.538, 95% confidence interval [CI] = 0.418–0.657), and that of the proportion of mesenchymal CTCs was 10.7% (AUC = 0.581, 95% CI = 0.463–0.699). We used the two cut-off values in combination to forecast PFS in which the total CTC count was equaled to or exceeded 10/5 mL with the proportion of mesenchymal CTCs surpassed 10.7%. Patients who met the combined criteria had significantly shorter median PFS than did those who did not meet the criteria (6.2 vs. 9.9 months, P =0.010). A nomogram was constructed based on the criteria and significant clinicopathological characteristics with a C-index of 0.613 (P = 0.010).ConclusionsThe criteria, which combine the total CTC count and the proportion of mesenchymal CTCs, may be used to monitor therapeutic resistance and predict prognosis in patients with metastatic breast cancer.Trial registration ClinicalTrials.gov. NCT01917279. Registered on 19 July 2013, https://clinicaltrials.gov/ct2/show/NCT01917279?term=NCT01917279&rank=1.

BackgroundTo characterize the clinical and pathological features and survival of patients with human epidermal growth factor receptor 2 (HER2)-low breast cancer in China.MethodsThe China National Cancer Center database was used to identify 1,433 metastatic breast cancer patients with HER2-negative disease diagnosed between 2005 and 2015. Clinicopathological features, survival, and prognosis information were extracted. Overall survival (OS) was estimated using the Kaplan–Meier method and compared using the log-rank test. Prognostic factors associated with OS were analyzed using Cox regression model with 95% confidence interval (95% CI).ResultsThere were 618 (43.1%) and 815 (56.9%) HER2-low and HER2-zero tumors out of 1,433 tumors, respectively. The proportion of hormone receptor (HR)-positive tumors was significantly higher in HER2-low tumors than in those with HER2-zero tumors (77.8% vs. 69.2%, p < 0.001). Patients with HER2-low tumors survived significantly longer than those with HER2-zero tumors in the overall population (48.5 months vs. 43.0 months, p = 0.004) and HR-positive subgroup (54.9 months vs. 48.1 months, p = 0.011), but not in the HR-negative subgroup (29.5 months vs. 29.9 months, p = 0.718). Multivariate regression analysis revealed that HER2-low tumors were independently associated with increased OS in HER2-negative population (HR: 0.85, 95% CI: 0.73–0.98, p = 0.026).ConclusionOur findings demonstrate that HER2-low tumors could be identified as a more distinct clinical entity from HER2-zero tumors, especially for the HR-positive subgroup. A more complex molecular landscape of HER2-low breast cancer might exist, and more precise diagnostic algorithms for HER2 testing could be investigated, thus offering new therapeutic targets for breast cancer treatment.

Purpose: This phase I study assessed the safety, tolerability, MTD, pharmacokinetics, antitumor activity, and predictive biomarkers of pyrotinib, an irreversible pan-ErbB inhibitor, in combination with capecitabine in patients with HER2positive metastatic breast cancer (MBC).Patients and Methods: Patients received oral pyrotinib 160 mg, 240 mg, 320 mg, or 400 mg once daily continually plus capecitabine 1,000 mg/m 2 twice daily on days 1 to 14 of a 21-day cycle. Pharmacokinetic blood samples were collected on days 1 and 14. Next-generation sequencing was performed on circulating tumor DNA to probe for predictive biomarkers.Results: A total of 28 patients were enrolled, 22 patients were treated at the two top-level doses. Among 17 (60.7%) trastuzumab-pretreated patients, 11 received trastuzumab for metastatic disease and 6 received adjuvant trastuzumab. No dose-limited toxicity was observed. Grade 3 treatment-related adverse events (AE) occurred in 12 (42.9%) patients; anemia (14.3%) and diarrhea (10.7%) were the most common grade 3 AEs. The overall response rate (ORR) was 78.6% [95% confidence interval (CI): 59.0%-91.7%], and the clinical benefit rate was 85.7% (95% CI: 67.3%-96.0%). The median progression-free survival (PFS) was 22.1 months (95% CI: 9.0-26.2 months). ORR was 70.6% (12/17) in trastuzumabpretreated patients and 90.9% (10/11) in trastuzumab-na€ ve patients. Analysis of all genetic alterations in HER2-related signaling network in baseline blood samples suggested that multiple genetic alterations were significantly associated with poorer PFS compared with none or one genetic alteration (median, 16.8 vs. 29.9 months, P ¼ 0.006).Conclusions: In a population largely na€ ve to HER2-targeted therapy, pyrotinib in combination with capecitabine was well-tolerated and demonstrates promising antitumor activity in patients with HER2-positive MBC.