- ⧉⌕$post array (20)
- ⇄datum => string (10) "18.05.2023"$post['datum']
- ⇄⧉veroeffentlichungsart => array (10)$post['veroeffentlichungsart']
- ⇄term_id => integer 838$post['veroeffentlichungsart']['term_id']
- ⇄name => UTF-8 string (22) "Zeitschriftenbeiträge"$post['veroeffentlichungsart']['name']
- ⇄slug => string (26) "246_zeitschriftenbeitraege"$post['veroeffentlichungsart']['slug']
- ⇄term_group => integer 0$post['veroeffentlichungsart']['term_group']
- ⇄term_taxonomy_id => integer 838$post['veroeffentlichungsart']['term_taxonomy_id']
- ⇄taxonomy => string (21) "veroeffentlichungsart"$post['veroeffentlichungsart']['taxonomy']
- ⇄description => string (0) ""$post['veroeffentlichungsart']['description']
- ⇄parent => integer 0$post['veroeffentlichungsart']['parent']
- ⇄count => integer 444$post['veroeffentlichungsart']['count']
- ⇄filter => string (3) "raw"$post['veroeffentlichungsart']['filter']
- ⇄term_id => integer 838
- ⇄⧉forschungsschwerpunkt => array (10)$post['forschungsschwerpunkt']
- ⇄term_id => integer 897$post['forschungsschwerpunkt']['term_id']
- ⇄name => UTF-8 string (51) "HRK Schwerpunkt Gesundheit analysieren und fördern"$post['forschungsschwerpunkt']['name']
- ⇄slug => string (55) "157_hrk-schwerpunkt-gesundheit-analysieren-und-foerdern"$post['forschungsschwerpunkt']['slug']
- ⇄term_group => integer 0$post['forschungsschwerpunkt']['term_group']
- ⇄term_taxonomy_id => integer 897$post['forschungsschwerpunkt']['term_taxonomy_id']
- ⇄taxonomy => string (21) "forschungsschwerpunkt"$post['forschungsschwerpunkt']['taxonomy']
- ⇄description => string (0) ""$post['forschungsschwerpunkt']['description']
- ⇄parent => integer 0$post['forschungsschwerpunkt']['parent']
- ⇄count => integer 83$post['forschungsschwerpunkt']['count']
- ⇄filter => string (3) "raw"$post['forschungsschwerpunkt']['filter']
- ⇄term_id => integer 897
- ⇄⧉autoren => array (17)$post['autoren']
- Table (17)
- Contents (17)
name link intern personenkennziffer 0 Lisa Schneider false 1 Roman Rischke https://www.hs-coburg.de/personen/prof-dr-roman-rischke/ true 2 Joachim Krois false 3 Aleksander Krasowski false 4 Martha Büttner false 5 Hossein Mohammad-Rahimi false 6 Akhilanand Chaurasia false 7 Nielsen S. Pereira false 8 Jae-Hong Lee false 9 Sergio E. Uribe false 10 Shahriar Shahab false 11 Revan B. Koca-Ünsal false 12 Gürkan Ünsal false 13 Yolanda Martinez-Beneyto false 14 Janet Brinz false 15 Olga Tryfonos false 16 Falk Schwendicke false - ⇄⧉0 => array (4)$post['autoren'][0]
- ⇄name => string (14) "Lisa Schneider"$post['autoren'][0]['name']
- ⇄link => string (0) ""$post['autoren'][0]['link']
- ⇄intern => boolean false$post['autoren'][0]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][0]['personenkennziffer']
- ⇄name => string (14) "Lisa Schneider"
- ⇄⧉1 => array (4)$post['autoren'][1]
- ⇄name => string (13) "Roman Rischke"$post['autoren'][1]['name']
- ⇄link => string (56) "https://www.hs-coburg.de/personen/prof-dr-roman-rischke/"$post['autoren'][1]['link']
- ⇄intern => boolean true$post['autoren'][1]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][1]['personenkennziffer']
- ⇄name => string (13) "Roman Rischke"
- ⇄⧉2 => array (4)$post['autoren'][2]
- ⇄name => string (13) "Joachim Krois"$post['autoren'][2]['name']
- ⇄link => string (0) ""$post['autoren'][2]['link']
- ⇄intern => boolean false$post['autoren'][2]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][2]['personenkennziffer']
- ⇄name => string (13) "Joachim Krois"
- ⇄⧉3 => array (4)$post['autoren'][3]
- ⇄name => string (20) "Aleksander Krasowski"$post['autoren'][3]['name']
- ⇄link => string (0) ""$post['autoren'][3]['link']
- ⇄intern => boolean false$post['autoren'][3]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][3]['personenkennziffer']
- ⇄name => string (20) "Aleksander Krasowski"
- ⇄⧉4 => array (4)$post['autoren'][4]
- ⇄name => UTF-8 string (15) "Martha Büttner"$post['autoren'][4]['name']
- ⇄link => string (0) ""$post['autoren'][4]['link']
- ⇄intern => boolean false$post['autoren'][4]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][4]['personenkennziffer']
- ⇄name => UTF-8 string (15) "Martha Büttner"
- ⇄⧉5 => array (4)$post['autoren'][5]
- ⇄name => string (23) "Hossein Mohammad-Rahimi"$post['autoren'][5]['name']
- ⇄link => string (0) ""$post['autoren'][5]['link']
- ⇄intern => boolean false$post['autoren'][5]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][5]['personenkennziffer']
- ⇄name => string (23) "Hossein Mohammad-Rahimi"
- ⇄⧉6 => array (4)$post['autoren'][6]
- ⇄name => string (20) "Akhilanand Chaurasia"$post['autoren'][6]['name']
- ⇄link => string (0) ""$post['autoren'][6]['link']
- ⇄intern => boolean false$post['autoren'][6]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][6]['personenkennziffer']
- ⇄name => string (20) "Akhilanand Chaurasia"
- ⇄⧉7 => array (4)$post['autoren'][7]
- ⇄name => string (18) "Nielsen S. Pereira"$post['autoren'][7]['name']
- ⇄link => string (0) ""$post['autoren'][7]['link']
- ⇄intern => boolean false$post['autoren'][7]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][7]['personenkennziffer']
- ⇄name => string (18) "Nielsen S. Pereira"
- ⇄⧉8 => array (4)$post['autoren'][8]
- ⇄name => string (12) "Jae-Hong Lee"$post['autoren'][8]['name']
- ⇄link => string (0) ""$post['autoren'][8]['link']
- ⇄intern => boolean false$post['autoren'][8]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][8]['personenkennziffer']
- ⇄name => string (12) "Jae-Hong Lee"
- ⇄⧉9 => array (4)$post['autoren'][9]
- ⇄name => string (15) "Sergio E. Uribe"$post['autoren'][9]['name']
- ⇄link => string (0) ""$post['autoren'][9]['link']
- ⇄intern => boolean false$post['autoren'][9]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][9]['personenkennziffer']
- ⇄name => string (15) "Sergio E. Uribe"
- ⇄⧉10 => array (4)$post['autoren'][10]
- ⇄name => string (15) "Shahriar Shahab"$post['autoren'][10]['name']
- ⇄link => string (0) ""$post['autoren'][10]['link']
- ⇄intern => boolean false$post['autoren'][10]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][10]['personenkennziffer']
- ⇄name => string (15) "Shahriar Shahab"
- ⇄⧉11 => array (4)$post['autoren'][11]
- ⇄name => UTF-8 string (20) "Revan B. Koca-Ünsal"$post['autoren'][11]['name']
- ⇄link => string (0) ""$post['autoren'][11]['link']
- ⇄intern => boolean false$post['autoren'][11]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][11]['personenkennziffer']
- ⇄name => UTF-8 string (20) "Revan B. Koca-Ünsal"
- ⇄⧉12 => array (4)$post['autoren'][12]
- ⇄name => UTF-8 string (14) "Gürkan Ünsal"$post['autoren'][12]['name']
- ⇄link => string (0) ""$post['autoren'][12]['link']
- ⇄intern => boolean false$post['autoren'][12]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][12]['personenkennziffer']
- ⇄name => UTF-8 string (14) "Gürkan Ünsal"
- ⇄⧉13 => array (4)$post['autoren'][13]
- ⇄name => string (24) "Yolanda Martinez-Beneyto"$post['autoren'][13]['name']
- ⇄link => string (0) ""$post['autoren'][13]['link']
- ⇄intern => boolean false$post['autoren'][13]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][13]['personenkennziffer']
- ⇄name => string (24) "Yolanda Martinez-Beneyto"
- ⇄⧉14 => array (4)$post['autoren'][14]
- ⇄name => string (11) "Janet Brinz"$post['autoren'][14]['name']
- ⇄link => string (0) ""$post['autoren'][14]['link']
- ⇄intern => boolean false$post['autoren'][14]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][14]['personenkennziffer']
- ⇄name => string (11) "Janet Brinz"
- ⇄⧉15 => array (4)$post['autoren'][15]
- ⇄name => string (13) "Olga Tryfonos"$post['autoren'][15]['name']
- ⇄link => string (0) ""$post['autoren'][15]['link']
- ⇄intern => boolean false$post['autoren'][15]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][15]['personenkennziffer']
- ⇄name => string (13) "Olga Tryfonos"
- ⇄⧉16 => array (4)$post['autoren'][16]
- ⇄name => string (16) "Falk Schwendicke"$post['autoren'][16]['name']
- ⇄link => string (0) ""$post['autoren'][16]['link']
- ⇄intern => boolean false$post['autoren'][16]['intern']
- ⇄personenkennziffer => string (0) ""$post['autoren'][16]['personenkennziffer']
- ⇄name => string (16) "Falk Schwendicke"
- ⇄⧉0 => array (4)
- ⇄⧉titel => string (90) "Federated vs local vs central deep learning of tooth segmentation on panoram...$post['titel']
Federated vs local vs central deep learning of tooth segmentation on panoramic radiographs
- ⇄medien => string (26) "Journal of Dental Research"$post['medien']
- ⇄doi => string (43) "https://doi.org/10.1016/j.jdent.2023.104556"$post['doi']
- ⇄weblink => string (0) ""$post['weblink']
- ⇄⧉abstract => string (1876) "<p><b>Objective</b><br>Federated Learning (FL) enables collaborative trainin...$post['abstract']
<p><b>Objective</b><br>Federated Learning (FL) enables collaborative training of artificial intelligence (AI) models from multiple data sources without directly sharing data. Due to the large amount of sensitive data in dentistry, FL may be particularly relevant for oral and dental research and applications. This study, for the first time, employed FL for a dental task, automated tooth segmentation on panoramic radiographs.<br><b>Methods</b><br>We employed a dataset of 4,177 panoramic radiographs collected from nine different centers (n = 143 to n = 1881 per center) across the globe and used FL to train a machine learning model for tooth segmentation. FL performance was compared against Local Learning (LL), i.e., training models on isolated data from each center (assuming data sharing not to be an option). Further, the performance gap to Central Learning (CL), i.e., training on centrally pooled data (based on data sharing agreements) was quantified. Generalizability of models was evaluated on a pooled test dataset from all centers.<br><b>Results</b><br>For 8 out of 9 centers, FL outperformed LL with statistical significance (p<0.05); only the center providing the largest amount of data FL did not have such an advantage. For generalizability, FL outperformed LL across all centers. CL surpassed both FL and LL for performance and generalizability.<br>Conclusion<br>If data pooling (for CL) is not feasible, FL is shown to be a useful alternative to train performant and, more importantly, generalizable deep learning models in dentistry, where data protection barriers are high.<br><b>Clinical Significance</b><br>This study proves the validity and utility of FL in the field of dentistry, which encourages researchers to adopt this method to improve the generalizability of dental AI models and ease their transition to the clinical environment.<br></p>
- ⇄heft => string (0) ""$post['heft']
- ⇄band => string (8) "2023/135"$post['band']
- ⇄artikelnummer => string (6) "104556"$post['artikelnummer']
- ⇄isbn => string (0) ""$post['isbn']
- ⇄herausgeber => string (0) ""$post['herausgeber']
- ⇄seiten => string (0) ""$post['seiten']
- ⇄open_access => null$post['open_access']
- ⇄peer_reviewed => boolean true$post['peer_reviewed']
- ⇄detailseite => boolean false$post['detailseite']
- ⇄⧉zitierung => UTF-8 string (503) "Schneider, Lisa; Rischke, Roman; Krois, Joachim; Krasowski, Aleksander; Bütt...$post['zitierung']
Schneider, Lisa; Rischke, Roman; Krois, Joachim; Krasowski, Aleksander; Büttner, Martha; Mohammad-Rahimi, Hossein; Chaurasia, Akhilanand; Pereira, Nielsen S.; Lee, Jae-Hong; Uribe, Sergio E.; Shahab, Shahriar; Koca-Ünsal, Revan B.; Ünsal, Gürkan; Martinez-Beneyto, Yolanda; Brinz, Janet; Tryfonos, Olga; Schwendicke, Falk (2023): Federated vs local vs central deep learning of tooth segmentation on panoramic radiographs. Journal of Dental Research 2023/135, 104556. DOI: 10.1016/j.jdent.2023.104556
- ⇄⧉permalink => string (133) "https://www.hs-coburg.de/publikation/4199-federated-vs-local-vs-central-deep...$post['permalink']
https://www.hs-coburg.de/publikation/4199-federated-vs-local-vs-central-deep-learning-of-tooth-segmentation-on-panoramic-radiographs/
- ⇄datum => string (10) "18.05.2023"
Federated vs local vs central deep learning of tooth segmentation on panoramic radiographs
Objective
Federated Learning (FL) enables collaborative training of artificial intelligence (AI) models from multiple data sources without directly sharing data. Due to the large amount of sensitive data in dentistry, FL may be particularly relevant for oral and dental research and applications. This study, for the first time, employed FL for a dental task, automated tooth segmentation on panoramic radiographs.
Methods
We employed a dataset of 4,177 panoramic radiographs collected from nine different centers (n = 143 to n = 1881 per center) across the globe and used FL to train a machine learning model for tooth segmentation. FL performance was compared against Local Learning (LL), i.e., training models on isolated data from each center (assuming data sharing not to be an option). Further, the performance gap to Central Learning (CL), i.e., training on centrally pooled data (based on data sharing agreements) was quantified. Generalizability of models was evaluated on a pooled test dataset from all centers.
Results
For 8 out of 9 centers, FL outperformed LL with statistical significance (p<0.05); only the center providing the largest amount of data FL did not have such an advantage. For generalizability, FL outperformed LL across all centers. CL surpassed both FL and LL for performance and generalizability.
Conclusion
If data pooling (for CL) is not feasible, FL is shown to be a useful alternative to train performant and, more importantly, generalizable deep learning models in dentistry, where data protection barriers are high.
Clinical Significance
This study proves the validity and utility of FL in the field of dentistry, which encourages researchers to adopt this method to improve the generalizability of dental AI models and ease their transition to the clinical environment.
Titel:
Veröffentlichungsdatum:
Publikationsart:
Forschungsschwerpunkt:
Medien:
DOI:
Weblink:
Heft:
Band:
Artikelnummer:
ISBN:
Autoren:
Lisa Schneider, Roman Rischke, Joachim Krois, Aleksander Krasowski, Martha Büttner, Hossein Mohammad-Rahimi, Akhilanand Chaurasia, Nielsen S. Pereira, Jae-Hong Lee, Sergio E. Uribe, Shahriar Shahab, Revan B. Koca-Ünsal, Gürkan Ünsal, Yolanda Martinez-Beneyto, Janet Brinz, Olga Tryfonos, Falk Schwendicke
Medien:
Herausgeber:
Seiten:
Open Access:
Peer reviewed:
Zitierung: