Explainable AI identifies diagnostic cells of genetic AML subtypes.
Explainable AI is deemed essential for clinical applications as it allows rationalizing model predictions, helping to build trust between clinicians and automated decision support tools. We developed an inherently explainable AI model for the classification of acute myeloid leukemia subtypes from bl...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Public Library of Science (PLoS)
2023-03-01
|
| Series: | PLOS Digital Health |
| Online Access: | https://doi.org/10.1371/journal.pdig.0000187 |
| _version_ | 1797696246763225088 |
|---|---|
| author | Matthias Hehr Ario Sadafi Christian Matek Peter Lienemann Christian Pohlkamp Torsten Haferlach Karsten Spiekermann Carsten Marr |
| author_facet | Matthias Hehr Ario Sadafi Christian Matek Peter Lienemann Christian Pohlkamp Torsten Haferlach Karsten Spiekermann Carsten Marr |
| author_sort | Matthias Hehr |
| collection | DOAJ |
| description | Explainable AI is deemed essential for clinical applications as it allows rationalizing model predictions, helping to build trust between clinicians and automated decision support tools. We developed an inherently explainable AI model for the classification of acute myeloid leukemia subtypes from blood smears and found that high-attention cells identified by the model coincide with those labeled as diagnostically relevant by human experts. Based on over 80,000 single white blood cell images from digitized blood smears of 129 patients diagnosed with one of four WHO-defined genetic AML subtypes and 60 healthy controls, we trained SCEMILA, a single-cell based explainable multiple instance learning algorithm. SCEMILA could perfectly discriminate between AML patients and healthy controls and detected the APL subtype with an F1 score of 0.86±0.05 (mean±s.d., 5-fold cross-validation). Analyzing a novel multi-attention module, we confirmed that our algorithm focused with high concordance on the same AML-specific cells as human experts do. Applied to classify single cells, it is able to highlight subtype specific cells and deconvolve the composition of a patient's blood smear without the need of single-cell annotation of the training data. Our large AML genetic subtype dataset is publicly available, and an interactive online tool facilitates the exploration of data and predictions. SCEMILA enables a comparison of algorithmic and expert decision criteria and can present a detailed analysis of individual patient data, paving the way to deploy AI in the routine diagnostics for identifying hematopoietic neoplasms. |
| first_indexed | 2024-03-12T03:23:39Z |
| format | Article |
| id | doaj.art-010031bc6f9549dd948d6f0be25758d1 |
| institution | Directory Open Access Journal |
| issn | 2767-3170 |
| language | English |
| last_indexed | 2024-03-12T03:23:39Z |
| publishDate | 2023-03-01 |
| publisher | Public Library of Science (PLoS) |
| record_format | Article |
| series | PLOS Digital Health |
| spelling | doaj.art-010031bc6f9549dd948d6f0be25758d12023-09-03T13:44:53ZengPublic Library of Science (PLoS)PLOS Digital Health2767-31702023-03-0123e000018710.1371/journal.pdig.0000187Explainable AI identifies diagnostic cells of genetic AML subtypes.Matthias HehrArio SadafiChristian MatekPeter LienemannChristian PohlkampTorsten HaferlachKarsten SpiekermannCarsten MarrExplainable AI is deemed essential for clinical applications as it allows rationalizing model predictions, helping to build trust between clinicians and automated decision support tools. We developed an inherently explainable AI model for the classification of acute myeloid leukemia subtypes from blood smears and found that high-attention cells identified by the model coincide with those labeled as diagnostically relevant by human experts. Based on over 80,000 single white blood cell images from digitized blood smears of 129 patients diagnosed with one of four WHO-defined genetic AML subtypes and 60 healthy controls, we trained SCEMILA, a single-cell based explainable multiple instance learning algorithm. SCEMILA could perfectly discriminate between AML patients and healthy controls and detected the APL subtype with an F1 score of 0.86±0.05 (mean±s.d., 5-fold cross-validation). Analyzing a novel multi-attention module, we confirmed that our algorithm focused with high concordance on the same AML-specific cells as human experts do. Applied to classify single cells, it is able to highlight subtype specific cells and deconvolve the composition of a patient's blood smear without the need of single-cell annotation of the training data. Our large AML genetic subtype dataset is publicly available, and an interactive online tool facilitates the exploration of data and predictions. SCEMILA enables a comparison of algorithmic and expert decision criteria and can present a detailed analysis of individual patient data, paving the way to deploy AI in the routine diagnostics for identifying hematopoietic neoplasms.https://doi.org/10.1371/journal.pdig.0000187 |
| spellingShingle | Matthias Hehr Ario Sadafi Christian Matek Peter Lienemann Christian Pohlkamp Torsten Haferlach Karsten Spiekermann Carsten Marr Explainable AI identifies diagnostic cells of genetic AML subtypes. PLOS Digital Health |
| title | Explainable AI identifies diagnostic cells of genetic AML subtypes. |
| title_full | Explainable AI identifies diagnostic cells of genetic AML subtypes. |
| title_fullStr | Explainable AI identifies diagnostic cells of genetic AML subtypes. |
| title_full_unstemmed | Explainable AI identifies diagnostic cells of genetic AML subtypes. |
| title_short | Explainable AI identifies diagnostic cells of genetic AML subtypes. |
| title_sort | explainable ai identifies diagnostic cells of genetic aml subtypes |
| url | https://doi.org/10.1371/journal.pdig.0000187 |
| work_keys_str_mv | AT matthiashehr explainableaiidentifiesdiagnosticcellsofgeneticamlsubtypes AT ariosadafi explainableaiidentifiesdiagnosticcellsofgeneticamlsubtypes AT christianmatek explainableaiidentifiesdiagnosticcellsofgeneticamlsubtypes AT peterlienemann explainableaiidentifiesdiagnosticcellsofgeneticamlsubtypes AT christianpohlkamp explainableaiidentifiesdiagnosticcellsofgeneticamlsubtypes AT torstenhaferlach explainableaiidentifiesdiagnosticcellsofgeneticamlsubtypes AT karstenspiekermann explainableaiidentifiesdiagnosticcellsofgeneticamlsubtypes AT carstenmarr explainableaiidentifiesdiagnosticcellsofgeneticamlsubtypes |