• Something wrong with this record ?

Fully automated imaging protocol independent system for pituitary adenoma segmentation: a convolutional neural network-based model on sparsely annotated MRI

M. Černý, J. Kybic, M. Májovský, V. Sedlák, K. Pirgl, E. Misiorzová, R. Lipina, D. Netuka

. 2023 ; 46 (1) : 116. [pub] 20230510

Language English Country Germany

Document type Journal Article

Persistent link   https://www.medvik.cz/link/bmc23011456

This study aims to develop a fully automated imaging protocol independent system for pituitary adenoma segmentation from magnetic resonance imaging (MRI) scans that can work without user interaction and evaluate its accuracy and utility for clinical applications. We trained two independent artificial neural networks on MRI scans of 394 patients. The scans were acquired according to various imaging protocols over the course of 11 years on 1.5T and 3T MRI systems. The segmentation model assigned a class label to each input pixel (pituitary adenoma, internal carotid artery, normal pituitary gland, background). The slice segmentation model classified slices as clinically relevant (structures of interest in slice) or irrelevant (anterior or posterior to sella turcica). We used MRI data of another 99 patients to evaluate the performance of the model during training. We validated the model on a prospective cohort of 28 patients, Dice coefficients of 0.910, 0.719, and 0.240 for tumour, internal carotid artery, and normal gland labels, respectively, were achieved. The slice selection model achieved 82.5% accuracy, 88.7% sensitivity, 76.7% specificity, and an AUC of 0.904. A human expert rated 71.4% of the segmentation results as accurate, 21.4% as slightly inaccurate, and 7.1% as coarsely inaccurate. Our model achieved good results comparable with recent works of other authors on the largest dataset to date and generalized well for various imaging protocols. We discussed future clinical applications, and their considerations. Models and frameworks for clinical use have yet to be developed and evaluated.

References provided by Crossref.org

000      
00000naa a2200000 a 4500
001      
bmc23011456
003      
CZ-PrNML
005      
20230815090522.0
007      
ta
008      
230718s2023 gw f 000 0|eng||
009      
AR
024    7_
$a 10.1007/s10143-023-02014-3 $2 doi
035    __
$a (PubMed)37162632
040    __
$a ABA008 $b cze $d ABA008 $e AACR2
041    0_
$a eng
044    __
$a gw
100    1_
$a Černý, Martin $u Department of Neurosurgery and Neurooncology, 1st Faculty of Medicine, Charles University, Central Military Hospital Prague, U Vojenské nemocnice 1200, 169 02, Praha 6, Czech Republic. dr.martin.cerny@gmail.com $u 1st Faculty of Medicine, Charles University Prague, Kateřinská 1660/32, 121 08, Praha 2, Czech Republic. dr.martin.cerny@gmail.com $1 https://orcid.org/0000000286010554 $7 xx0304973
245    10
$a Fully automated imaging protocol independent system for pituitary adenoma segmentation: a convolutional neural network-based model on sparsely annotated MRI / $c M. Černý, J. Kybic, M. Májovský, V. Sedlák, K. Pirgl, E. Misiorzová, R. Lipina, D. Netuka
520    9_
$a This study aims to develop a fully automated imaging protocol independent system for pituitary adenoma segmentation from magnetic resonance imaging (MRI) scans that can work without user interaction and evaluate its accuracy and utility for clinical applications. We trained two independent artificial neural networks on MRI scans of 394 patients. The scans were acquired according to various imaging protocols over the course of 11 years on 1.5T and 3T MRI systems. The segmentation model assigned a class label to each input pixel (pituitary adenoma, internal carotid artery, normal pituitary gland, background). The slice segmentation model classified slices as clinically relevant (structures of interest in slice) or irrelevant (anterior or posterior to sella turcica). We used MRI data of another 99 patients to evaluate the performance of the model during training. We validated the model on a prospective cohort of 28 patients, Dice coefficients of 0.910, 0.719, and 0.240 for tumour, internal carotid artery, and normal gland labels, respectively, were achieved. The slice selection model achieved 82.5% accuracy, 88.7% sensitivity, 76.7% specificity, and an AUC of 0.904. A human expert rated 71.4% of the segmentation results as accurate, 21.4% as slightly inaccurate, and 7.1% as coarsely inaccurate. Our model achieved good results comparable with recent works of other authors on the largest dataset to date and generalized well for various imaging protocols. We discussed future clinical applications, and their considerations. Models and frameworks for clinical use have yet to be developed and evaluated.
650    _2
$a lidé $7 D006801
650    12
$a nádory hypofýzy $x diagnostické zobrazování $x chirurgie $7 D010911
650    _2
$a prospektivní studie $7 D011446
650    _2
$a magnetická rezonanční tomografie $7 D008279
650    _2
$a neuronové sítě $7 D016571
650    12
$a adenom $x diagnostické zobrazování $x chirurgie $7 D000236
650    _2
$a počítačové zpracování obrazu $x metody $7 D007091
655    _2
$a časopisecké články $7 D016428
700    1_
$a Kybic, Jan $u Department of Cybernetics, Faculty of Electrical Engineering, Czech Technical University in Prague, Technická 2, 166 27, Praha 6, Czech Republic $1 https://orcid.org/0000000293634947 $7 xx0028484
700    1_
$a Májovský, Martin $u Department of Neurosurgery and Neurooncology, 1st Faculty of Medicine, Charles University, Central Military Hospital Prague, U Vojenské nemocnice 1200, 169 02, Praha 6, Czech Republic $1 https://orcid.org/0000000177255181 $7 xx0228525
700    1_
$a Sedlák, Vojtěch $u Department of Radiodiagnostics, Central Military Hospital Prague, U Vojenské nemocnice 1200, 169 02, Praha 6, Czech Republic $1 https://orcid.org/0000000170319680
700    1_
$a Pirgl, Karin $u Department of Neurosurgery and Neurooncology, 1st Faculty of Medicine, Charles University, Central Military Hospital Prague, U Vojenské nemocnice 1200, 169 02, Praha 6, Czech Republic $u 3rd Faculty of Medicine, Charles University Prague, Ruská 87, 100 00, Praha 10, Czech Republic $1 https://orcid.org/0000000208262799
700    1_
$a Misiorzová, Eva $u Department of Neurosurgery, Faculty of Medicine, University of Ostrava, University Hospital Ostrava, 17. listopadu 1790/5, 708 52, Ostrava-Poruba, Czech Republic $1 https://orcid.org/0000000229937900
700    1_
$a Lipina, Radim $u Department of Neurosurgery, Faculty of Medicine, University of Ostrava, University Hospital Ostrava, 17. listopadu 1790/5, 708 52, Ostrava-Poruba, Czech Republic $1 https://orcid.org/0000000283881311
700    1_
$a Netuka, David $u Department of Neurosurgery and Neurooncology, 1st Faculty of Medicine, Charles University, Central Military Hospital Prague, U Vojenské nemocnice 1200, 169 02, Praha 6, Czech Republic $1 https://orcid.org/0000000186094789 $7 xx0061783
773    0_
$w MED00003513 $t Neurosurgical review $x 1437-2320 $g Roč. 46, č. 1 (2023), s. 116
856    41
$u https://pubmed.ncbi.nlm.nih.gov/37162632 $y Pubmed
910    __
$a ABA008 $b sig $c sign $y p $z 0
990    __
$a 20230718 $b ABA008
991    __
$a 20230815090519 $b ABA008
999    __
$a ok $b bmc $g 1963706 $s 1197721
BAS    __
$a 3
BAS    __
$a PreBMC-MEDLINE
BMC    __
$a 2023 $b 46 $c 1 $d 116 $e 20230510 $i 1437-2320 $m Neurosurgical review $n Neurosurg Rev $x MED00003513
LZP    __
$a Pubmed-20230718

Find record

Citation metrics

Archiving options