1
|
Cheng N, Ren Y, Zhou J, Zhang Y, Wang D, Zhang X, Chen B, Liu F, Lv J, Cao Q, Chen S, Du H, Hui D, Weng Z, Liang Q, Su B, Tang L, Han L, Chen J, Shao C. Deep Learning-Based Classification of Hepatocellular Nodular Lesions on Whole-Slide Histopathologic Images. Gastroenterology 2022; 162:1948-1961.e7. [PMID: 35202643 DOI: 10.1053/j.gastro.2022.02.025] [Citation(s) in RCA: 32] [Impact Index Per Article: 10.7] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 05/01/2021] [Revised: 12/21/2021] [Accepted: 02/15/2022] [Indexed: 12/11/2022]
Abstract
BACKGROUND & AIMS Hepatocellular nodular lesions (HNLs) constitute a heterogeneous group of disorders. Differential diagnosis among these lesions, especially high-grade dysplastic nodules (HGDNs) and well-differentiated hepatocellular carcinoma (WD-HCC), can be challenging, let alone biopsy specimens. We aimed to develop a deep learning system to solve these puzzles, improving the histopathologic diagnosis of HNLs (WD-HCC, HGDN, low-grade DN, focal nodular hyperplasia, hepatocellular adenoma), and background tissues (nodular cirrhosis, normal liver tissue). METHODS The samples consisting of surgical and biopsy specimens were collected from 6 hospitals. Each specimen was reviewed by 2 to 3 subspecialists. Four deep neural networks (ResNet50, InceptionV3, Xception, and the Ensemble) were used. Their performances were evaluated by confusion matrix, receiver operating characteristic curve, classification map, and heat map. The predictive efficiency of the optimal model was further verified by comparing with that of 9 pathologists. RESULTS We obtained 213,280 patches from 1115 whole-slide images of 738 patients. An optimal model was finally chosen based on F1 score and area under the curve value, named hepatocellular-nodular artificial intelligence model (HnAIM), with the overall 7-category area under the curve of 0.935 in the independent external validation cohort. For biopsy specimens, the agreement rate with subspecialists' majority opinion was higher for HnAIM than 9 pathologists on both patch level and whole-slide images level. CONCLUSIONS We first developed a deep learning diagnostic model for HNLs, which performed well and contributed to enhancing the diagnosis rate of early HCC and risk stratification of patients with HNLs. Furthermore, HnAIM had significant advantages in patch-level recognition, with important diagnostic implications for fragmentary or scarce biopsy specimens.
Collapse
Affiliation(s)
- Na Cheng
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Yong Ren
- Digestive Diseases Center, The Seventh Affiliated Hospital, Sun Yat-sen University, Shenzhen, China; Center for Artificial Intelligence in Medicine, Research Institute of Tsinghua, Pearl River Delta, Guangzhou, China
| | - Jing Zhou
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Yiwang Zhang
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Deyu Wang
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Xiaofang Zhang
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Bing Chen
- Department of Pathology, The Third Affiliated Hospital of Sun Yat-sen University Yuedong Hospital, Meizhou, China
| | - Fang Liu
- Department of Pathology, FoShan First People's Hospital, Foshan, China
| | - Jin Lv
- Department of Pathology, FoShan First People's Hospital, Foshan, China
| | - Qinghua Cao
- Department of Pathology, The First Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Sijin Chen
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Hong Du
- Department of Pathology, GuangZhou First People's Hospital, Guangzhou, China
| | - Dayang Hui
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Zijin Weng
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Qiong Liang
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Bojin Su
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
| | - Luying Tang
- Department of Pathology, The Third Affiliated Hospital of Sun Yat-sen University Lingnan Hospital, Guangzhou, China
| | - Lanqing Han
- Center for Artificial Intelligence in Medicine, Research Institute of Tsinghua, Pearl River Delta, Guangzhou, China.
| | - Jianning Chen
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China.
| | - Chunkui Shao
- Department of Pathology, The Third Affiliated Hospital, Sun Yat-sen University, Guangzhou, China.
| |
Collapse
|
2
|
Sheng X, Qin JM. Clinical features and diagnostic and therapeutic strategies of hepatic dysplastic nodules. Shijie Huaren Xiaohua Zazhi 2022; 30:169-181. [DOI: 10.11569/wcjd.v30.i4.169] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 02/06/2023] Open
Abstract
Hepatic dysplastic nodules (DNs) are a group of neoplastic lesions with a diameter of more than 1 cm that belong to precancerous lesions, with abnormal cytoplasm and cells but without malignant basis in histology. Hepatic DNs lack typical tumor markers and clinical symptoms, and their clinical diagnosis relys mainly on imaging or/and tissue pathological examination. Thanks to the further research on the pathogenesis of hepatic DNs and the development of imaging technology, the combination of medical history, various examinationss, individual tumor markers, and imaging and histopathology techniques can significantly improve the early detection and diagnosis accuracy for hepatic DNs, and reduce the rate of missed and false diagnosis. Due to the potential malignancy risk of hepatic DNs, intervention measures should be carried out on hepatic DNs at all stages, in order to block the transformation process of DNs into hepatocellular carcinoma (HCC), which is of great clinical significance to reduce the incidence and mortality of HCC.
Collapse
Affiliation(s)
- Xia Sheng
- Department of Pathology, Minhang Hospital Affiliated to Fudan University, Shanghai 201100, China
| | - Jian-Min Qin
- Department of General Surgery, The Third Hospital Affiliated to Naval Military Medical University, Shanghai 201805, China
| |
Collapse
|
3
|
Bioulac-Sage P, Gouw ASH, Balabaud C, Sempoux C. Hepatocellular Adenoma: What We Know, What We Do Not Know, and Why It Matters. Histopathology 2021; 80:878-897. [PMID: 34856012 DOI: 10.1111/his.14605] [Citation(s) in RCA: 21] [Impact Index Per Article: 5.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/05/2021] [Revised: 11/21/2021] [Accepted: 11/25/2021] [Indexed: 11/27/2022]
Abstract
In the last 2 decades there has been significant progress in research and diagnosis of hepatocellular adenoma (HCA), resulting in the establishment of a molecular and immunohistological HCA classification. This review aims to fine-tune the current expertise in order to enhance the histopathological diagnostic possibilities, by refining issues that are already known, addressing diagnostic difficulties and identifying still unknown aspects of HCA. We will discuss novel methods to identify HCA subtypes, in particular the sonic hedgehog HCAs and the interpretation of glutamine synthetase patterns for the recognition of beta-catenin mutated HCAs. The major complications of HCAs, bleeding and malignant transformation, will be considered, including the dilemmas of atypical and borderline lesions. Paragraphs on HCAs in different clinical and geographical settings, e.g. pregnancy, cirrhosis and non-western countries are included. The natural history of the different HCA subtypes in relation with age, sex and risk factors is a feature still insufficiently investigated. This is also true for the risks of clinical bleeding and malignant transformation in association with HCA subtypes. As HCA is a relatively rare tumor, a multicenter and multidisciplinary approach across geographical boundaries will be the appropriate method to establish prospective programs to identify, classify and manage HCAs, focusing on several aspects, e.g. etiology, underlying liver disease, complications, regression and growth. Updating what we know, identifying and addressing features that we do not know matters to warrant optimal patient management.
Collapse
Affiliation(s)
| | - Annette S H Gouw
- Departement of Pathology and Medical Biology, University Medical Center Groningen, Groningen, the Netherlands
| | | | - Christine Sempoux
- Service of Clinical Pathology, Institute of Pathology, Lausanne University Hospital and University of Lausanne, Lausanne, Switzerland
| |
Collapse
|
4
|
Zhu M, Li H, Wang C, Yang B, Wang X, Hou F, Yang S, Wang Y, Guo X, Qi X. Focal nodular hyperplasia mimicking hepatocellular adenoma and carcinoma in two cases. Drug Discov Ther 2021; 15:112-117. [PMID: 33952777 DOI: 10.5582/ddt.2021.01033] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/07/2023]
Abstract
Focal nodular hyperplasia (FNH) is a solid benign tumor of the liver, predominantly in young women. A correct diagnosis of FNH is essential for making appropriate clinical decisions and avoiding unnecessary liver resection. Herein, we reported that two male cases with FNH, who initially presented with persistent abdominal discomfort, were misdiagnosed with hepatocellular adenoma (HCA) and hepatocellular carcinoma (HCC) on contrast-enhanced magnetic resonance imaging and computed tomography scans, respectively. After surgery, a histological diagnosis of FNH was finally established. In this paper, we also reviewed the knowledge regarding diagnosis and differential diagnosis of FNH on imaging examinations, which are helpful for avoiding misdiagnoses and guiding clinical interventions.
Collapse
Affiliation(s)
- Menghua Zhu
- Department of Gastroenterology, General Hospital of Northern Theater Command, Shenyang, China
- Postgraduate College, Jinzhou Medical University, Jinzhou, China
| | - Hongyu Li
- Department of Gastroenterology, General Hospital of Northern Theater Command, Shenyang, China
| | - Chunhui Wang
- Department of Hepatobiliary Surgery, General Hospital of Northern Theater Command, Shenyang, China
| | - Benqiang Yang
- Department of Radiology, General Hospital of Northern Theater Command, Shenyang, China
| | - Xuehan Wang
- Department of Pathology, General Hospital of Northern Theater Command, Shenyang, China
| | - Feifei Hou
- Department of Gastroenterology, General Hospital of Northern Theater Command, Shenyang, China
| | - Shengye Yang
- Department of Gastroenterology, General Hospital of Northern Theater Command, Shenyang, China
| | - Yuye Wang
- Department of Gastroenterology, General Hospital of Northern Theater Command, Shenyang, China
- Postgraduate College, Jinzhou Medical University, Jinzhou, China
| | - Xiaozhong Guo
- Department of Gastroenterology, General Hospital of Northern Theater Command, Shenyang, China
| | - Xingshun Qi
- Department of Gastroenterology, General Hospital of Northern Theater Command, Shenyang, China
| |
Collapse
|
5
|
Saito K, Ledsam J, Sourbron S, Araki Y. Validation study of perfusion parameter in hypervascular hepatocellular carcinoma and focal nodular hyperplasia using dynamic susceptibility magnetic resonance imaging with super-paramagnetic iron oxide: comparison with single level dynamic CT arteriography. Quant Imaging Med Surg 2020; 10:1298-1306. [PMID: 32550138 DOI: 10.21037/qims-18-233] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.2] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/14/2022]
Abstract
Background Dynamic susceptibility contrast MR imaging (DSC-MRI) offers direct evaluation of neo-vascularity. Ferucarbotran does not accumulate in the interstitial space, instead remaining in the intravascular space during early phase imaging. We investigate tracer kinetic analysis with DSC-MRI with ferucarbotran and single level CT during hepatic arteriography (SL-CTHA) in assessment of hypervascular hepatocellular lesions and evaluate the usefulness of DSC-MRI with ferucarbotran. Methods Six patients having hypervascular hepatocellular carcinoma (HCC) and 3 patients having focal nodular hyperplasia (FNH) were included in the study. SL-CTHA was performed with the infusion of 3 mL of contrast media at a rate of 1 mL/s and scanned at a rate of 0.8 second per rotation. DSC-MRI was acquired with the echo-planar method at 1.5T system. A total dose of 1.4 mL (0.5 mol Fe/L) of ferucarbotran was used. Ferucarbotran was injected at a rate of 2 mL/s with 40 mL of physiological saline. Imaging was obtained at a temporal resolution of 1.2 or 0.46 seconds in 5 and 4 patients, respectively. For both CT and MRI modalities, a model-free analysis method was used to derive region of interest-based perfusion parameters. Plasma flow, distribution volume (DV) of contrast agent and estimated mean transit time (EMTT) were estimated. Results A strong correlation was obtained with plasma flow (r=0.8231, P=0.0064) between DSC-MRI and SL-CTHA. No significant correlation was obtained for DV and EMTT between DSC-MRI and SL-CTHA. All perfusion parameters showed no significant difference between SL-CTHA and DSC-MRI in FNH. On the other hand, in HCC, DV and EMTT showed significant differences (P=0.046 and 0.046), and plasma flow showed no significant difference between DSC-MRI and SL-CTHA. Conclusions This pilot study demonstrates the possibility of quantitative analysis of liver tumor using superparamagnetic iron oxide (SPIO)-based agent and highlights the potential for SPIO-based agent in more precisely assessing the perfusion characteristic of hypervascular liver tumors than by using extracellular contrast media.
Collapse
Affiliation(s)
- Kazuhiro Saito
- Department of Radiology, Tokyo Medical University, Shinjuku-ku, Tokyo, Japan
| | - Joseph Ledsam
- Division of Biomedical Imaging, Leeds University, Leeds, UK
| | | | - Yoichi Araki
- Department of Radiology, Tokyo Medical University, Shinjuku-ku, Tokyo, Japan
| |
Collapse
|
6
|
Akhaladze D, Grachev N, Kachanov D, Talypov S, Merkulov N, Uskova N, Andreev E, Rabaev G, Ivanova N, Varfolomeeva S. Liver resection for atypical giant focal nodular hyperplasia. JOURNAL OF PEDIATRIC SURGERY CASE REPORTS 2019. [DOI: 10.1016/j.epsc.2019.101202] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/27/2022] Open
|
7
|
Tsilimigras DI, Rahnemai-Azar AA, Ntanasis-Stathopoulos I, Gavriatopoulou M, Moris D, Spartalis E, Cloyd JM, Weber SM, Pawlik TM. Current Approaches in the Management of Hepatic Adenomas. J Gastrointest Surg 2019; 23:199-209. [PMID: 30109469 DOI: 10.1007/s11605-018-3917-4] [Citation(s) in RCA: 18] [Impact Index Per Article: 3.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 07/04/2018] [Accepted: 08/03/2018] [Indexed: 02/08/2023]
Abstract
INTRODUCTION Hepatic adenomas (HAs) are a benign and relatively rare type of liver neoplasms. We review the diagnosis, evaluation, and potential therapeutic management options for patients with HA. METHODS A comprehensive review of the English literature was performed utilizing MEDLINE/PubMed and Web of Science databases with end of search date the 30th April of 2018. In PubMed, the terms "hepatocellular," "hepatic," "liver," and "adenoma," "adenomatosis" were searched in the title and/or abstract. RESULTS Recent advances in molecular classification of HA have determined distinct subtypes with specific clinical, pathological, and imaging characteristics. In general, cessation of exogenous hormonal administration or weight loss may lead to HA regression. Surgical resection, either open or laparoscopic, should be considered in patients with symptoms and risk factors for hemorrhage or malignant transformation. These risk factors include tumor diameter greater than 5 cm, β-catenin activated subtype, and/or male gender. The management of acute hemorrhage should primarily aim at achieving hemodynamic stability via angioembolization followed by elective resection, whereas malignant transformation is treated according to oncologic resection principles. Although pregnancy is one of the known risk factors for tumor growth and associated complications, the presence of an HA per se should not be considered a contradiction to pregnancy. CONCLUSION Future genomic-based multicenter studies are required to provide a strong basis for formulating an evidence-based risk-adapted model that guides individualized management strategies for patients with HA.
Collapse
Affiliation(s)
- Diamantis I Tsilimigras
- Laboratory of Experimental Surgery and Surgical Research, School of Medicine, National and Kapodistrian University of Athens, Athens, Greece
| | - Amir A Rahnemai-Azar
- Department of Surgery, Division of Surgical Oncology, University of Wisconsin Hospital, Madison, WI, USA
| | - Ioannis Ntanasis-Stathopoulos
- Department of Clinical Therapeutics, School of Medicine, National and Kapodistrian University of Athens, Alexandra General Hospital, Athens, Greece
| | - Maria Gavriatopoulou
- Department of Clinical Therapeutics, School of Medicine, National and Kapodistrian University of Athens, Alexandra General Hospital, Athens, Greece
| | - Demetrios Moris
- Department of Surgery, Division of Surgical Oncology, The Ohio State University Wexner Medical Center and James Cancer Hospital and Solove Research Institute, Columbus, OH, USA
| | - Eleftherios Spartalis
- Laboratory of Experimental Surgery and Surgical Research, School of Medicine, National and Kapodistrian University of Athens, Athens, Greece
| | - Jordan M Cloyd
- Department of Surgery, Division of Surgical Oncology, The Ohio State University Wexner Medical Center and James Cancer Hospital and Solove Research Institute, Columbus, OH, USA
| | - Sharon M Weber
- Department of Surgery, Division of Surgical Oncology, University of Wisconsin Hospital, Madison, WI, USA
| | - Timothy M Pawlik
- Department of Surgery, Division of Surgical Oncology, The Ohio State University Wexner Medical Center and James Cancer Hospital and Solove Research Institute, Columbus, OH, USA. .,Department of Surgery, The Urban Meyer III and Shelley Meyer Chair for Cancer Research, Oncology, Health Services Management and Policy, The Ohio State University, Wexner Medical Center, 395 W. 12th Ave., Suite 670, Columbus, OH, USA.
| |
Collapse
|
8
|
Growth of Focal Nodular Hyperplasia is Not a Reason for Surgical Intervention, but Patients Should be Referred to a Tertiary Referral Centre. World J Surg 2018; 42:1506-1513. [PMID: 29167952 PMCID: PMC5895671 DOI: 10.1007/s00268-017-4335-6] [Citation(s) in RCA: 7] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 12/31/2022]
Abstract
Background When a liver lesion diagnosed as focal nodular hyperplasia (FNH) increases in size, it may cause doubt about the initial diagnosis. In many cases, additional investigations will follow to exclude hepatocellular adenoma or malignancy. This retrospective cohort study addresses the implications of growth of FNH for clinical management. Methods We included patients diagnosed with FNH based on ≥2 imaging modalities between 2002 and 2015. Characteristics of patients with growing FNH with sequential imaging in a 6-month interval were compared to non-growing FNH. Results Growth was reported in 19/162 (12%) patients, ranging from 21 to 200%. Resection was performed in 4/19 growing FNHs; histological examination confirmed FNH in all patients. In all 15 conservatively treated patients, additional imaging confirmed FNH diagnosis. No adverse outcomes were reported. No differences were found in characteristics and presentation of patients with growing or non-growing FNH. Conclusion This study confirms that FNH may grow significantly without causing symptoms. A significant increase in size should not have any implications on clinical management if confident diagnosis by imaging has been established by a tertiary benign liver multidisciplinary team. Liver biopsy is only indicated in case of doubt after state-of-the-art imaging. Resection is deemed unnecessary if the diagnosis is confirmed by multiple imaging modalities in a tertiary referral centre.
Collapse
|
9
|
Sempoux C, Balabaud C, Paradis V, Bioulac-Sage P. Hepatocellular nodules in vascular liver diseases. Virchows Arch 2018; 473:33-44. [PMID: 29804132 DOI: 10.1007/s00428-018-2373-6] [Citation(s) in RCA: 25] [Impact Index Per Article: 3.6] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/21/2018] [Revised: 05/03/2018] [Accepted: 05/08/2018] [Indexed: 12/14/2022]
Abstract
Hepatocellular nodules have been recognized in vascular liver diseases for a long time and mostly described and studied in the imaging literature. Some confusions in their identification and overlap in their definitions exist, especially in this specific clinical context. Pathology descriptions report the development of nodular regenerative hyperplasia, large regenerative nodule, and focal nodular hyperplasia, as adaptive responses of the liver parenchyma to the modified blood flow. True neoplastic hepatocellular nodules such as hepatocellular adenoma and hepatocellular carcinoma can also appear, mainly in Budd-Chiari syndrome, and have to be correctly diagnosed. This is more difficult for the radiologist in these diseased livers, leading more frequently to perform liver biopsies. We describe the histology of each type of well-differentiated hepatocellular nodules and provide some clues for their differential diagnosis. A review of the literature gives an historical perspective of the problem and enlightens the frequency and the subtypes of hepatocellular nodules found in the most common vascular liver diseases.
Collapse
Affiliation(s)
- Christine Sempoux
- Service of Clinical Pathology, Lausanne University Hospital, Institute of Pathology, 25, rue du Bugnon, 1011, Lausanne, Switzerland.
| | - Charles Balabaud
- Inserm, UMR1053 Bordeaux Research In Translational Oncology, BaRITOn, F-33000 Bordeaux, Université de Bordeaux, 33076, Bordeaux, France
| | - Valérie Paradis
- Pathology department, Beaujon hospital, Inserm UMR 1149, Université Paris Diderot, 101 bd du Général Leclerc, 92110, Clichy, France
| | - Paulette Bioulac-Sage
- Inserm, UMR1053 Bordeaux Research In Translational Oncology, BaRITOn, F-33000 Bordeaux, Université de Bordeaux, 33076, Bordeaux, France
| |
Collapse
|