Predrag Janjic1, Kristijan Petrovski2, Blagoja Dolgoski3, John Smiley4, Panche Zdravkovski3, Goran Pavlovski5, Zlatko Jakjovski5, Natasa Davceva5, Verica Poposka5, Aleksandar Stankov5, Gorazd Rosoklija6, Gordana Petrushevska3, Ljupco Kocarev7, Andrew J Dwork8. 1. Research Center for Computer Science and Information Technology, Macedonian Academy of Sciences and Arts, Bul. Krste Misirkov 2, 1000, Skopje, North Macedonia. Electronic address: predrag.a.janjic@gmail.com. 2. Research Center for Computer Science and Information Technology, Macedonian Academy of Sciences and Arts, Bul. Krste Misirkov 2, 1000, Skopje, North Macedonia. 3. Institute of Pathology, School of Medicine, Ss. Cyril and Methodius University Skopje, ul. 50ta Divizija 6, 1000, Skopje, North Macedonia. 4. Nathan S. Kline Institute for Psychiatric Research, 140 Old Orangeburg Road, Orangeburg, NY 10962, USA. 5. Institute of Forensic Medicine, School of Medicine, Ss. Cyril and Methodius University Skopje, ul. 50ta Divizija 6, 1000, Skopje, North Macedonia. 6. Department of Psychiatry, Columbia University, New York, USA; Division of Molecular Imaging and Neuropathology, New York State Psychiatric Institute, 1051 Riverside Drive, Unit 42, New York, NY, 10032, USA; Macedonian Academy of Sciences and Arts, Bul. Krste Misirkov 2, 1000, Skopje, North Macedonia. 7. Research Center for Computer Science and Information Technology, Macedonian Academy of Sciences and Arts, Bul. Krste Misirkov 2, 1000, Skopje, North Macedonia; Faculty of Computer Science and Engineering, Ss. Cyril and Methodius University, ul. Rudjer Boskovic 16, PO Box 393, Skopje, North Macedonia. 8. Department of Psychiatry, Columbia University, New York, USA; Division of Molecular Imaging and Neuropathology, New York State Psychiatric Institute, 1051 Riverside Drive, Unit 42, New York, NY, 10032, USA; Macedonian Academy of Sciences and Arts, Bul. Krste Misirkov 2, 1000, Skopje, North Macedonia; Department of Pathology and Cell Biology, Columbia University, New York, USA.
Abstract
BACKGROUND: Standard segmentation of high-contrast electron micrographs (EM) identifies myelin accurately but does not translate easily into measurements of individual axons and their myelin, even in cross-sections of parallel fibers. We describe automated segmentation and measurement of each myelinated axon and its sheath in EMs of arbitrarily oriented human white matter from autopsies. NEW METHODS: Preliminary segmentation of myelin, axons and background by machine learning, using selected filters, precedes automated correction of systematic errors. Final segmentation is done by a deep neural network (DNN). Automated measurement of each putative fiber rejects measures encountering pre-defined artifacts and excludes fibers failing to satisfy pre-defined conditions. RESULTS: Improved segmentation of three sets of 30 annotated images each (two sets from human prefrontal white matter and one from human optic nerve) is achieved with a DNN trained only with a subset of the first set from prefrontal white matter. Total number of myelinated axons identified by the DNN differed from expert segmentation by 0.2%, 2.9%, and -5.1%, respectively. G-ratios differed by 2.96%, 0.74% and 2.83%. Intraclass correlation coefficients between DNN and annotated segmentation were mostly >0.9, indicating nearly interchangeable performance. COMPARISON WITH EXISTING METHOD(S): Measurement-oriented studies of arbitrarily oriented fibers from central white matter are rare. Published methods are typically applied to cross-sections of fascicles and measure aggregated areas of myelin sheaths and axons, allowing estimation only of average g-ratio. CONCLUSIONS: Automated segmentation and measurement of axons and myelin is complex. We report a feasible approach that has so far proven comparable to manual segmentation.
BACKGROUND: Standard segmentation of high-contrast electron micrographs (EM) identifies myelin accurately but does not translate easily into measurements of individual axons and their myelin, even in cross-sections of parallel fibers. We describe automated segmentation and measurement of each myelinated axon and its sheath in EMs of arbitrarily oriented human white matter from autopsies. NEW METHODS: Preliminary segmentation of myelin, axons and background by machine learning, using selected filters, precedes automated correction of systematic errors. Final segmentation is done by a deep neural network (DNN). Automated measurement of each putative fiber rejects measures encountering pre-defined artifacts and excludes fibers failing to satisfy pre-defined conditions. RESULTS: Improved segmentation of three sets of 30 annotated images each (two sets from human prefrontal white matter and one from human optic nerve) is achieved with a DNN trained only with a subset of the first set from prefrontal white matter. Total number of myelinated axons identified by the DNN differed from expert segmentation by 0.2%, 2.9%, and -5.1%, respectively. G-ratios differed by 2.96%, 0.74% and 2.83%. Intraclass correlation coefficients between DNN and annotated segmentation were mostly >0.9, indicating nearly interchangeable performance. COMPARISON WITH EXISTING METHOD(S): Measurement-oriented studies of arbitrarily oriented fibers from central white matter are rare. Published methods are typically applied to cross-sections of fascicles and measure aggregated areas of myelin sheaths and axons, allowing estimation only of average g-ratio. CONCLUSIONS: Automated segmentation and measurement of axons and myelin is complex. We report a feasible approach that has so far proven comparable to manual segmentation.
Authors: Bastian G Brinkmann; Amit Agarwal; Michael W Sereda; Alistair N Garratt; Thomas Müller; Hagen Wende; Ruth M Stassart; Schanila Nawaz; Christian Humml; Viktorija Velanac; Konstantin Radyushkin; Sandra Goebbels; Tobias M Fischer; Robin J Franklin; Cary Lai; Hannelore Ehrenreich; Carmen Birchmeier; Markus H Schwab; Klaus Armin Nave Journal: Neuron Date: 2008-08-28 Impact factor: 17.173
Authors: Galin V Michailov; Michael W Sereda; Bastian G Brinkmann; Tobias M Fischer; Bernhard Haug; Carmen Birchmeier; Lorna Role; Cary Lai; Markus H Schwab; Klaus-Armin Nave Journal: Science Date: 2004-03-25 Impact factor: 47.728