Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline
The manual observation of sputum smears by fluorescence microscopy for the diagnosis and treatment monitoring of patients with tuberculosis (TB) is a laborious and subjective task. In this work, we introduce an automatic pipeline which employs a novel deep learning-based approach to rapidly detect M...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-02-01
|
| Series: | Information |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2078-2489/13/2/96 |
| _version_ | 1797479224022401024 |
|---|---|
| author | Marios Zachariou Ognjen Arandjelović Wilber Sabiiti Bariki Mtafya Derek Sloan |
| author_facet | Marios Zachariou Ognjen Arandjelović Wilber Sabiiti Bariki Mtafya Derek Sloan |
| author_sort | Marios Zachariou |
| collection | DOAJ |
| description | The manual observation of sputum smears by fluorescence microscopy for the diagnosis and treatment monitoring of patients with tuberculosis (TB) is a laborious and subjective task. In this work, we introduce an automatic pipeline which employs a novel deep learning-based approach to rapidly detect Mycobacterium tuberculosis (Mtb) organisms in sputum samples and thus quantify the burden of the disease. Fluorescence microscopy images are used as input in a series of networks, which ultimately produces a final count of present bacteria more quickly and consistently than manual analysis by healthcare workers. The pipeline consists of four stages: annotation by cycle-consistent generative adversarial networks (GANs), extraction of salient image patches, classification of the extracted patches, and finally, regression to yield the final bacteria count. We empirically evaluate the individual stages of the pipeline as well as perform a unified evaluation on previously unseen data that were given ground-truth labels by an experienced microscopist. We show that with no human intervention, the pipeline can provide the bacterial count for a sample of images with an error of less than 5%. |
| first_indexed | 2024-03-09T21:42:45Z |
| format | Article |
| id | doaj.art-10a39c5cb03f42de809eafe65b9bbfb7 |
| institution | Directory Open Access Journal |
| issn | 2078-2489 |
| language | English |
| last_indexed | 2024-03-09T21:42:45Z |
| publishDate | 2022-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Information |
| spelling | doaj.art-10a39c5cb03f42de809eafe65b9bbfb72023-11-23T20:25:47ZengMDPI AGInformation2078-24892022-02-011329610.3390/info13020096Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning PipelineMarios Zachariou0Ognjen Arandjelović1Wilber Sabiiti2Bariki Mtafya3Derek Sloan4School of Computer Science, University of St Andrews, St Andrews KY16 9AJ, UKSchool of Computer Science, University of St Andrews, St Andrews KY16 9AJ, UKSchool of Medicine, University of St Andrews, St Andrews KY16 9TF, UKMbeya Medical Research Center, Mbeya 2410, TanzaniaSchool of Medicine, University of St Andrews, St Andrews KY16 9TF, UKThe manual observation of sputum smears by fluorescence microscopy for the diagnosis and treatment monitoring of patients with tuberculosis (TB) is a laborious and subjective task. In this work, we introduce an automatic pipeline which employs a novel deep learning-based approach to rapidly detect Mycobacterium tuberculosis (Mtb) organisms in sputum samples and thus quantify the burden of the disease. Fluorescence microscopy images are used as input in a series of networks, which ultimately produces a final count of present bacteria more quickly and consistently than manual analysis by healthcare workers. The pipeline consists of four stages: annotation by cycle-consistent generative adversarial networks (GANs), extraction of salient image patches, classification of the extracted patches, and finally, regression to yield the final bacteria count. We empirically evaluate the individual stages of the pipeline as well as perform a unified evaluation on previously unseen data that were given ground-truth labels by an experienced microscopist. We show that with no human intervention, the pipeline can provide the bacterial count for a sample of images with an error of less than 5%.https://www.mdpi.com/2078-2489/13/2/96cycle GANssemantic segmentationpatch extractionsaliencyclassificationregression |
| spellingShingle | Marios Zachariou Ognjen Arandjelović Wilber Sabiiti Bariki Mtafya Derek Sloan Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline Information cycle GANs semantic segmentation patch extraction saliency classification regression |
| title | Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline |
| title_full | Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline |
| title_fullStr | Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline |
| title_full_unstemmed | Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline |
| title_short | Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline |
| title_sort | tuberculosis bacteria detection and counting in fluorescence microscopy images using a multi stage deep learning pipeline |
| topic | cycle GANs semantic segmentation patch extraction saliency classification regression |
| url | https://www.mdpi.com/2078-2489/13/2/96 |
| work_keys_str_mv | AT marioszachariou tuberculosisbacteriadetectionandcountinginfluorescencemicroscopyimagesusingamultistagedeeplearningpipeline AT ognjenarandjelovic tuberculosisbacteriadetectionandcountinginfluorescencemicroscopyimagesusingamultistagedeeplearningpipeline AT wilbersabiiti tuberculosisbacteriadetectionandcountinginfluorescencemicroscopyimagesusingamultistagedeeplearningpipeline AT barikimtafya tuberculosisbacteriadetectionandcountinginfluorescencemicroscopyimagesusingamultistagedeeplearningpipeline AT dereksloan tuberculosisbacteriadetectionandcountinginfluorescencemicroscopyimagesusingamultistagedeeplearningpipeline |