Deep learning based identification of bone scintigraphies containing metastatic bone disease foci
Abstract Purpose Metastatic bone disease (MBD) is the most common form of metastases, most frequently deriving from prostate cancer. MBD is screened with bone scintigraphy (BS), which have high sensitivity but low specificity for the diagnosis of MBD, often requiring further investigations. Deep lea...
| Main Authors: | , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2023-01-01
|
| Series: | Cancer Imaging |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s40644-023-00524-3 |
| _version_ | 1811175772658860032 |
|---|---|
| author | Abdalla Ibrahim Akshayaa Vaidyanathan Sergey Primakov Flore Belmans Fabio Bottari Turkey Refaee Pierre Lovinfosse Alexandre Jadoul Celine Derwael Fabian Hertel Henry C. Woodruff Helle D. Zacho Sean Walsh Wim Vos Mariaelena Occhipinti François-Xavier Hanin Philippe Lambin Felix M. Mottaghy Roland Hustinx |
| author_facet | Abdalla Ibrahim Akshayaa Vaidyanathan Sergey Primakov Flore Belmans Fabio Bottari Turkey Refaee Pierre Lovinfosse Alexandre Jadoul Celine Derwael Fabian Hertel Henry C. Woodruff Helle D. Zacho Sean Walsh Wim Vos Mariaelena Occhipinti François-Xavier Hanin Philippe Lambin Felix M. Mottaghy Roland Hustinx |
| author_sort | Abdalla Ibrahim |
| collection | DOAJ |
| description | Abstract Purpose Metastatic bone disease (MBD) is the most common form of metastases, most frequently deriving from prostate cancer. MBD is screened with bone scintigraphy (BS), which have high sensitivity but low specificity for the diagnosis of MBD, often requiring further investigations. Deep learning (DL) - a machine learning technique designed to mimic human neuronal interactions- has shown promise in the field of medical imaging analysis for different purposes, including segmentation and classification of lesions. In this study, we aim to develop a DL algorithm that can classify areas of increased uptake on bone scintigraphy scans. Methods We collected 2365 BS from three European medical centres. The model was trained and validated on 1203 and 164 BS scans respectively. Furthermore we evaluated its performance on an external testing set composed of 998 BS scans. We further aimed to enhance the explainability of our developed algorithm, using activation maps. We compared the performance of our algorithm to that of 6 nuclear medicine physicians. Results The developed DL based algorithm is able to detect MBD on BSs, with high specificity and sensitivity (0.80 and 0.82 respectively on the external test set), in a shorter time compared to the nuclear medicine physicians (2.5 min for AI and 30 min for nuclear medicine physicians to classify 134 BSs). Further prospective validation is required before the algorithm can be used in the clinic. |
| first_indexed | 2024-04-10T19:41:20Z |
| format | Article |
| id | doaj.art-c744ea8c106646c39197f1afa9bedf77 |
| institution | Directory Open Access Journal |
| issn | 1470-7330 |
| language | English |
| last_indexed | 2024-04-10T19:41:20Z |
| publishDate | 2023-01-01 |
| publisher | BMC |
| record_format | Article |
| series | Cancer Imaging |
| spelling | doaj.art-c744ea8c106646c39197f1afa9bedf772023-01-29T12:20:53ZengBMCCancer Imaging1470-73302023-01-012311910.1186/s40644-023-00524-3Deep learning based identification of bone scintigraphies containing metastatic bone disease fociAbdalla Ibrahim0Akshayaa Vaidyanathan1Sergey Primakov2Flore Belmans3Fabio Bottari4Turkey Refaee5Pierre Lovinfosse6Alexandre Jadoul7Celine Derwael8Fabian Hertel9Henry C. Woodruff10Helle D. Zacho11Sean Walsh12Wim Vos13Mariaelena Occhipinti14François-Xavier Hanin15Philippe Lambin16Felix M. Mottaghy17Roland Hustinx18The D-Lab, Department of Precision Medicine, GROW- School for Oncology and Developmental Biology, Maastricht UniversityThe D-Lab, Department of Precision Medicine, GROW- School for Oncology and Developmental Biology, Maastricht UniversityThe D-Lab, Department of Precision Medicine, GROW- School for Oncology and Developmental Biology, Maastricht UniversityRadiomics (Oncoradiomics SA)Radiomics (Oncoradiomics SA)The D-Lab, Department of Precision Medicine, GROW- School for Oncology and Developmental Biology, Maastricht UniversityDivision of Nuclear Medicine and Oncological Imaging, Department of Medical Physics, University Hospital of LiegeDivision of Nuclear Medicine and Oncological Imaging, Department of Medical Physics, University Hospital of LiegeDivision of Nuclear Medicine and Oncological Imaging, Department of Medical Physics, University Hospital of LiegeDepartment of Nuclear Medicine and Comprehensive diagnostic centre Aachen (CDCA), University Hospital RWTH Aachen UniversityThe D-Lab, Department of Precision Medicine, GROW- School for Oncology and Developmental Biology, Maastricht UniversityDepartment of Nuclear Medicine, Clinical Cancer Research Centre, Aalborg University HospitalRadiomics (Oncoradiomics SA)Radiomics (Oncoradiomics SA)Radiomics (Oncoradiomics SA)Department of Nuclear Medicine, Universite´CatholiqueUniversite´Catholique de Louvain, CHU-UCL-NamurThe D-Lab, Department of Precision Medicine, GROW- School for Oncology and Developmental Biology, Maastricht UniversityDepartment of Radiology and Nuclear Medicine, Columbia University Irving Medical CenterDivision of Nuclear Medicine and Oncological Imaging, Department of Medical Physics, University Hospital of LiegeAbstract Purpose Metastatic bone disease (MBD) is the most common form of metastases, most frequently deriving from prostate cancer. MBD is screened with bone scintigraphy (BS), which have high sensitivity but low specificity for the diagnosis of MBD, often requiring further investigations. Deep learning (DL) - a machine learning technique designed to mimic human neuronal interactions- has shown promise in the field of medical imaging analysis for different purposes, including segmentation and classification of lesions. In this study, we aim to develop a DL algorithm that can classify areas of increased uptake on bone scintigraphy scans. Methods We collected 2365 BS from three European medical centres. The model was trained and validated on 1203 and 164 BS scans respectively. Furthermore we evaluated its performance on an external testing set composed of 998 BS scans. We further aimed to enhance the explainability of our developed algorithm, using activation maps. We compared the performance of our algorithm to that of 6 nuclear medicine physicians. Results The developed DL based algorithm is able to detect MBD on BSs, with high specificity and sensitivity (0.80 and 0.82 respectively on the external test set), in a shorter time compared to the nuclear medicine physicians (2.5 min for AI and 30 min for nuclear medicine physicians to classify 134 BSs). Further prospective validation is required before the algorithm can be used in the clinic.https://doi.org/10.1186/s40644-023-00524-3Deep learningMetastatic bone diseaseBone scintigraphyActivation maps |
| spellingShingle | Abdalla Ibrahim Akshayaa Vaidyanathan Sergey Primakov Flore Belmans Fabio Bottari Turkey Refaee Pierre Lovinfosse Alexandre Jadoul Celine Derwael Fabian Hertel Henry C. Woodruff Helle D. Zacho Sean Walsh Wim Vos Mariaelena Occhipinti François-Xavier Hanin Philippe Lambin Felix M. Mottaghy Roland Hustinx Deep learning based identification of bone scintigraphies containing metastatic bone disease foci Cancer Imaging Deep learning Metastatic bone disease Bone scintigraphy Activation maps |
| title | Deep learning based identification of bone scintigraphies containing metastatic bone disease foci |
| title_full | Deep learning based identification of bone scintigraphies containing metastatic bone disease foci |
| title_fullStr | Deep learning based identification of bone scintigraphies containing metastatic bone disease foci |
| title_full_unstemmed | Deep learning based identification of bone scintigraphies containing metastatic bone disease foci |
| title_short | Deep learning based identification of bone scintigraphies containing metastatic bone disease foci |
| title_sort | deep learning based identification of bone scintigraphies containing metastatic bone disease foci |
| topic | Deep learning Metastatic bone disease Bone scintigraphy Activation maps |
| url | https://doi.org/10.1186/s40644-023-00524-3 |
| work_keys_str_mv | AT abdallaibrahim deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT akshayaavaidyanathan deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT sergeyprimakov deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT florebelmans deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT fabiobottari deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT turkeyrefaee deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT pierrelovinfosse deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT alexandrejadoul deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT celinederwael deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT fabianhertel deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT henrycwoodruff deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT helledzacho deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT seanwalsh deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT wimvos deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT mariaelenaocchipinti deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT francoisxavierhanin deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT philippelambin deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT felixmmottaghy deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci AT rolandhustinx deeplearningbasedidentificationofbonescintigraphiescontainingmetastaticbonediseasefoci |