Dataset Information

Computer-assisted mitotic count using a deep learning-based algorithm improves interobserver reproducibility and accuracy.

ABSTRACT: The mitotic count (MC) is an important histological parameter for prognostication of malignant neoplasms. However, it has inter- and intraobserver discrepancies due to difficulties in selecting the region of interest (MC-ROI) and in identifying or classifying mitotic figures (MFs). Recent progress in the field of artificial intelligence has allowed the development of high-performance algorithms that may improve standardization of the MC. As algorithmic predictions are not flawless, computer-assisted review by pathologists may ensure reliability. In the present study, we compared partial (MC-ROI preselection) and full (additional visualization of MF candidates and display of algorithmic confidence values) computer-assisted MC analysis to the routine (unaided) MC analysis by 23 pathologists for whole-slide images of 50 canine cutaneous mast cell tumors (ccMCTs). Algorithmic predictions aimed to assist pathologists in detecting mitotic hotspot locations, reducing omission of MFs, and improving classification against imposters. The interobserver consistency for the MC significantly increased with computer assistance (interobserver correlation coefficient, ICC = 0.92) compared to the unaided approach (ICC = 0.70). Classification into prognostic stratifications had a higher accuracy with computer assistance. The algorithmically preselected hotspot MC-ROIs had a consistently higher MCs than the manually selected MC-ROIs. Compared to a ground truth (developed with immunohistochemistry for phosphohistone H3), pathologist performance in detecting individual MF was augmented when using computer assistance (F1-score of 0.68 increased to 0.79) with a reduction in false negatives by 38%. The results of this study demonstrate that computer assistance may lead to more reproducible and accurate MCs in ccMCTs.

SUBMITTER: Bertram CA

PROVIDER: S-EPMC8928234 | biostudies-literature | 2022 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Computer-assisted mitotic count using a deep learning-based algorithm improves interobserver reproducibility and accuracy.

Bertram Christof A CA Aubreville Marc M Donovan Taryn A TA Bartel Alexander A Wilm Frauke F Marzahl Christian C Assenmacher Charles-Antoine CA Becker Kathrin K Bennett Mark M Corner Sarah S Cossic Brieuc B Denk Daniela D Dettwiler Martina M Gonzalez Beatriz Garcia BG Gurtner Corinne C Haverkamp Ann-Kathrin AK Heier Annabelle A Lehmbecker Annika A Merz Sophie S Noland Erica L EL Plog Stephanie S Schmidt Anja A Sebastian Franziska F Sledge Dodd G DG Smedley Rebecca C RC Tecilla Marco M Thaiwong Tuddow T Fuchs-Baumgartinger Andrea A Meuten Donald J DJ Breininger Katharina K Kiupel Matti M Maier Andreas A Klopfleisch Robert R

Veterinary pathology 20211230 2

The mitotic count (MC) is an important histological parameter for prognostication of malignant neoplasms. However, it has inter- and intraobserver discrepancies due to difficulties in selecting the region of interest (MC-ROI) and in identifying or classifying mitotic figures (MFs). Recent progress in the field of artificial intelligence has allowed the development of high-performance algorithms that may improve standardization of the MC. As algorithmic predictions are not flawless, computer-assi ...[more]

PMID: 34965805

Similar Datasets

Project description:BackgroundIncreased accuracy and lower rates of component positioning outliers have been associated with better long-term survival and functional outcomes of total knee arthroplasty (TKA). This study investigates the accuracy of robotic-assisted TKA compared to navigation-assisted and manual instrumentation techniques, using polyethylene tibial insert thickness as a surrogate.MethodsConsecutive primary TKA by a single surgeon were retrospectively reviewed and divided in 3 groups: manual instrumentation, navigation-assisted, and robotic-assisted (RA-TKA). Polyethylene insert thickness, deviation from planned thickness, and rate of outliers were compared between the 3 groups using nonparametric analysis of variance, Kruskal-Wallis tests, and Bonferroni corrections. Logistic regression analysis was performed to identify predictors of polyethylene thickness ≥9 mm. The learning curve for RA-TKA was evaluated with a box plot graph of groups of 10 consecutive cases.ResultsThere were 474 patients in manual instrumentation TKA, 257 in navigation-assisted TKA and 225 in RA-TKA, with median polyethylene thicknesses of 6.0 (interquartile range 5.0-7.0), 6.0 (interquartile range 5.0-7.0), and 5.0 (interquartile range 5.0-6.0) millimeters, respectively (P˂0.001 RA-TKA compared to both other groups). Polyethylene inserts with a thickness ≥9 mm were used in 28 (5.9%) manual instrumentation TKA, 13 (5.1%) navigation-assisted TKA, and 1 (0.4%) RA-TKA (P = .004). Independent predictors for polyethylene thickness ≥9 mm included surgical technique, left side, and male sex. A learning curve of <30 cases was observed before consistent polyethylene thickness was achieved in RA-TKA.ConclusionsTibial polyethylene insert thickness, as a surrogate of surgical accuracy, is more reproducible in robotic-assisted than in navigation-assisted or manual-instrumentation TKA. The learning curve to reach high levels of reproducibility with this technique is relatively short.

Dataset Information

Computer-assisted mitotic count using a deep learning-based algorithm improves interobserver reproducibility and accuracy.

Publications

Computer-assisted mitotic count using a deep learning-based algorithm improves interobserver reproducibility and accuracy.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets