1
|
Ibrahim A, Jahanifar M, Wahab N, Toss MS, Makhlouf S, Atallah N, Lashen AG, Katayama A, Graham S, Bilal M, Bhalerao A, Ahmed Raza SE, Snead D, Minhas F, Rajpoot N, Rakha E. Artificial Intelligence-Based Mitosis Scoring in Breast Cancer: Clinical Application. Mod Pathol 2024; 37:100416. [PMID: 38154653 DOI: 10.1016/j.modpat.2023.100416] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/16/2023] [Revised: 10/27/2023] [Accepted: 12/14/2023] [Indexed: 12/30/2023]
Abstract
In recent years, artificial intelligence (AI) has demonstrated exceptional performance in mitosis identification and quantification. However, the implementation of AI in clinical practice needs to be evaluated against the existing methods. This study is aimed at assessing the optimal method of using AI-based mitotic figure scoring in breast cancer (BC). We utilized whole slide images from a large cohort of BC with extended follow-up comprising a discovery (n = 1715) and a validation (n = 859) set (Nottingham cohort). The Cancer Genome Atlas of breast invasive carcinoma (TCGA-BRCA) cohort (n = 757) was used as an external test set. Employing automated mitosis detection, the mitotic count was assessed using 3 different methods, the mitotic count per tumor area (MCT; calculated by dividing the number of mitotic figures by the total tumor area), the mitotic index (MI; defined as the average number of mitotic figures per 1000 malignant cells), and the mitotic activity index (MAI; defined as the number of mitotic figures in 3 mm2 area within the mitotic hotspot). These automated metrics were evaluated and compared based on their correlation with the well-established visual scoring method of the Nottingham grading system and Ki67 score, clinicopathologic parameters, and patient outcomes. AI-based mitotic scores derived from the 3 methods (MCT, MI, and MAI) were significantly correlated with the clinicopathologic characteristics and patient survival (P < .001). However, the mitotic counts and the derived cutoffs varied significantly between the 3 methods. Only MAI and MCT were positively correlated with the gold standard visual scoring method used in Nottingham grading system (r = 0.8 and r = 0.7, respectively) and Ki67 scores (r = 0.69 and r = 0.55, respectively), and MAI was the only independent predictor of survival (P < .05) in multivariate Cox regression analysis. For clinical applications, the optimum method of scoring mitosis using AI needs to be considered. MAI can provide reliable and reproducible results and can accurately quantify mitotic figures in BC.
Collapse
Affiliation(s)
- Asmaa Ibrahim
- Academic Unit for Translational Medical Sciences, School of Medicine, University of Nottingham, Nottingham, United Kingdom; Department of Pathology, Faculty of Medicine, Suez Canal University, Egypt
| | - Mostafa Jahanifar
- Tissue Image Analytics Centre, University of Warwick, United Kingdom
| | - Noorul Wahab
- Tissue Image Analytics Centre, University of Warwick, United Kingdom
| | - Michael S Toss
- Academic Unit for Translational Medical Sciences, School of Medicine, University of Nottingham, Nottingham, United Kingdom; Histopathology Department, Royal Hallamshire Hospital, Sheffield Teaching Hospitals NHS Foundation Trust, Sheffield, United Kingdom
| | - Shorouk Makhlouf
- Academic Unit for Translational Medical Sciences, School of Medicine, University of Nottingham, Nottingham, United Kingdom
| | - Nehal Atallah
- Academic Unit for Translational Medical Sciences, School of Medicine, University of Nottingham, Nottingham, United Kingdom
| | - Ayat G Lashen
- Academic Unit for Translational Medical Sciences, School of Medicine, University of Nottingham, Nottingham, United Kingdom
| | - Ayaka Katayama
- Academic Unit for Translational Medical Sciences, School of Medicine, University of Nottingham, Nottingham, United Kingdom
| | - Simon Graham
- Tissue Image Analytics Centre, University of Warwick, United Kingdom
| | - Mohsin Bilal
- Tissue Image Analytics Centre, University of Warwick, United Kingdom
| | - Abhir Bhalerao
- Tissue Image Analytics Centre, University of Warwick, United Kingdom
| | - Shan E Ahmed Raza
- Tissue Image Analytics Centre, University of Warwick, United Kingdom
| | - David Snead
- Cellular Pathology, University Hospitals Coventry and Warwickshire NHS Trust, United Kingdom
| | - Fayyaz Minhas
- Tissue Image Analytics Centre, University of Warwick, United Kingdom
| | - Nasir Rajpoot
- Tissue Image Analytics Centre, University of Warwick, United Kingdom.
| | - Emad Rakha
- Academic Unit for Translational Medical Sciences, School of Medicine, University of Nottingham, Nottingham, United Kingdom; Nottingham University Hospitals NHS Trust, Nottingham, United Kingdom; Pathology Department, Hamad Medical Corporation, Doha, Qatar.
| |
Collapse
|
2
|
Renne SL. How to measure your microscope's HPF. A critical guide for residents. Pathologica 2023; 115:302-307. [PMID: 38180138 PMCID: PMC10767800 DOI: 10.32074/1591-951x-900] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/13/2023] [Accepted: 10/10/2023] [Indexed: 01/06/2024] Open
Abstract
Counting stuff under the microscope is part of the duties of a surgical pathologist. Many textbooks and articles still report the surface area as the number of high-power fields (HPFs) counted. This is bad, since the area displayed by an HPF varies between two microscopes. It is therefore necessary to express the surface as mm2. This is a how to guide written for the resident who has to measure the HPF of the microscope for the first time. The Resident can either calibrate the microscope with a stage micrometer slide (a small ruler on a glass slide) or compute the surface area of the HPF using the numbers on the eyepiece and the magnification objective. for "10X/22" eyepiece and a "40X" objective, the diameter of the HPF is 22/40 = 0.55 (if no other magnification is present), and the surface is 0.238 mm2. The young resident might then ask: "How far off-target was I when I counted the number of HPFs that the chief resident declared to be correct?" Probably not that much: although legitimate in principle and correct in math, the size of the problem is often overstated since microscopes are not that different after all and because pathology is not just about counting.
Collapse
Affiliation(s)
- Salvatore Lorenzo Renne
- Department of Biomedical Sciences, Humanitas University, Milan, Italy
- IRCCS Humanitas Research Hospital, Rozzano, Milan, Italy
| |
Collapse
|
3
|
A Novel Deep Learning-Based Mitosis Recognition Approach and Dataset for Uterine Leiomyosarcoma Histopathology. Cancers (Basel) 2022; 14:cancers14153785. [PMID: 35954449 PMCID: PMC9367529 DOI: 10.3390/cancers14153785] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/18/2022] [Revised: 07/24/2022] [Accepted: 08/01/2022] [Indexed: 11/17/2022] Open
Abstract
Uterine leiomyosarcoma (ULMS) is the most common sarcoma of the uterus, with both a high malignant potential and poor prognosis. Its diagnosis is sometimes challenging owing to its resemblance to leiomyosarcoma, often being accompanied by benign smooth muscle neoplasms of the uterus. Pathologists diagnose and grade leiomyosarcoma based on three biomarkers (i.e., mitosis count, necrosis, and nuclear atypia). Among these biomarkers, mitosis count is the most important and challenging biomarker. In general, pathologists use the traditional manual counting method for the detection and counting of mitosis. This procedure is very time-consuming, tedious, and subjective. To overcome these challenges, artificial intelligence (AI) based methods have been developed that automatically detect mitosis. In this paper, we propose a new ULMS dataset and an AI-based approach for mitosis detection. We collected our dataset from a local medical facility in collaboration with highly trained pathologists. Preprocessing and annotations are performed using standard procedures, and a deep learning-based method is applied to provide baseline accuracies. The experimental results showed 0.7462 precision, 0.8981 recall, and 0.8151 F1-score. For research and development, the code and dataset have been made publicly available.
Collapse
|
4
|
Walker RJB, Look Hong NJ, Moncrieff M, van Akkooi ACJ, Jost E, Nessim C, van Houdt WJ, Stahlie EHA, Seo C, Quan ML, McKinnon JG, Wright FC, Mavros MN. Predictors of Sentinel Lymph Node Metastasis in Patients with Thin Melanoma: An International Multi-institutional Collaboration. Ann Surg Oncol 2022; 29:7010-7017. [PMID: 35676603 DOI: 10.1245/s10434-022-11936-z] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/12/2022] [Accepted: 05/10/2022] [Indexed: 11/18/2022]
Abstract
BACKGROUND Consideration of sentinel lymph node biopsy (SLNB) is recommended for patients with T1b melanomas and T1a melanomas with high-risk features; however, the proportion of patients with actionable results is low. We aimed to identify factors predicting SLNB positivity in T1 melanomas by examining a multi-institutional international population. METHODS Data were extracted on patients with T1 cutaneous melanoma who underwent SLNB between 2005 and 2018 at five tertiary centers in Europe and Canada. Univariable and multivariable logistic regression analyses were performed to identify predictors of SLNB positivity. RESULTS Overall, 676 patients were analyzed. Most patients had one or more high-risk features: Breslow thickness 0.8-1 mm in 78.1% of patients, ulceration in 8.3%, mitotic rate > 1/mm2 in 42.5%, Clark's level ≥ 4 in 34.3%, lymphovascular invasion in 1.4%, nodular histology in 2.9%, and absence of tumor-infiltrating lymphocytes in 14.4%. Fifty-three patients (7.8%) had a positive SLNB. Breslow thickness and mitotic rate independently predicted SLNB positivity. The odds of positive SLNB increased by 50% for each 0.1 mm increase in thickness past 0.7 mm (95% confidence interval [CI] 1.05-2.13) and by 22% for each mitosis per mm2 (95% CI 1.06-1.41). Patients who had one excised node (vs. two or more) were three times less likely to have a positive SLNB (3.6% vs. 9.6%; odds ratio 2.9 [1.3-7.7]). CONCLUSIONS Our international multi-institutional data confirm that Breslow thickness and mitotic rate independently predict SLNB positivity in patients with T1 melanoma. Even within this highly selected population, the number needed to diagnose is 13:1 (7.8%), indicating that more work is required to identify additional predictors of sentinel node positivity.
Collapse
Affiliation(s)
- Richard J B Walker
- Department of Surgery, Sunnybrook Health Sciences Centre, Toronto, ON, Canada
| | - Nicole J Look Hong
- Department of Surgery, Sunnybrook Health Sciences Centre, Toronto, ON, Canada
| | - Marc Moncrieff
- Department of Plastic & Reconstructive Surgery, Norfolk and Norwich University Hospital, Norwich, UK
| | - Alexander C J van Akkooi
- Melanoma Institute Australia, The University of Sydney and Royal Prince Alfred Hospital, Sidney, Australia
| | - Evan Jost
- Department of Surgery, Foothills Medical Centre, Calgary, AB, Canada
| | - Carolyn Nessim
- Department of Surgery, The Ottawa Hospital, OHRI, Ottawa, ON, Canada
| | - Winan J van Houdt
- Department of Surgery, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, The Netherlands
| | - Emma H A Stahlie
- Department of Surgery, Netherlands Cancer Institute - Antoni van Leeuwenhoek, Amsterdam, The Netherlands
| | - Chanhee Seo
- Department of Surgery, The Ottawa Hospital, OHRI, Ottawa, ON, Canada
| | - May Lynn Quan
- Department of Surgery, Foothills Medical Centre, Calgary, AB, Canada
| | | | - Frances C Wright
- Department of Surgery, Sunnybrook Health Sciences Centre, Toronto, ON, Canada
| | - Michail N Mavros
- Department of Surgery, University of Arkansas for Medical Sciences, Little Rock, AR, USA.
| |
Collapse
|
5
|
Cree IA, Tan PH, Travis WD, Wesseling P, Yagi Y, White VA, Lokuhetty D, Scolyer RA. Counting mitoses: SI(ze) matters! Mod Pathol 2021; 34:1651-1657. [PMID: 34079071 PMCID: PMC8376633 DOI: 10.1038/s41379-021-00825-7] [Citation(s) in RCA: 59] [Impact Index Per Article: 19.7] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/16/2021] [Revised: 04/16/2021] [Accepted: 04/20/2021] [Indexed: 11/08/2022]
Abstract
Mitoses are often assessed by pathologists to assist the diagnosis of cancer, and to grade malignancy, informing prognosis. Historically, this has been done by expressing the number of mitoses per n high power fields (HPFs), ignoring the fact that microscope fields may differ substantially, even at the same high power (×400) magnification. Despite a requirement to define HPF size in scientific papers, many authors fail to address this issue adequately. The problem is compounded by the switch to digital pathology systems, where ×400 equivalent fields are rectangular and also vary in the area displayed. The potential for error is considerable, and at times this may affect patient care. This is easily solved by the use of standardized international (SI) units. We, therefore, recommend that features such as mitoses are always counted per mm2, with an indication of the area to be counted and the method used (usually "hotspot" or "average") to obtain the results.
Collapse
Affiliation(s)
- Ian A Cree
- International Agency for Research on Cancer (IARC), World Health Organization, Lyon, France.
| | - Puay Hoon Tan
- Division of Pathology, Singapore General Hospital, Singapore, Singapore
| | - William D Travis
- Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, NY, USA
| | - Pieter Wesseling
- Department of Pathology, Amsterdam University Medical Centers/VUmc, Amsterdam, The Netherlands
- Princess Máxima Center for Pediatric Oncology, Utrecht, The Netherlands
| | - Yukako Yagi
- Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, NY, USA
| | - Valerie A White
- International Agency for Research on Cancer (IARC), World Health Organization, Lyon, France
| | - Dilani Lokuhetty
- International Agency for Research on Cancer (IARC), World Health Organization, Lyon, France
- Department of Pathology, Faculty of Medicine, University of Colombo, Colombo, Sri Lanka
| | - Richard A Scolyer
- Melanoma Institute Australia and Faculty of Medicine and Health, The University of Sydney, Tissue Pathology and Diagnostic Oncology, Royal Prince Alfred Hospital, and NSW Health Pathology, Sydney, NSW, Australia
| |
Collapse
|