Investigating recurrent neural networks for OCT A-scan based tissue analysis

TY - JOUR

T1 - Investigating recurrent neural networks for OCT A-scan based tissue analysis

AU - Otte, C

AU - Otte, S

AU - Wittig, L

AU - Hüttmann, G

AU - Kugler, C

AU - Drömann, D

AU - Zell, A

AU - Schlaefer, A

PY - 2014

Y1 - 2014

N2 - OBJECTIVES: Optical Coherence Tomography (OCT) has been proposed as a high resolution image modality to guide transbronchial biopsies. In this study we address the question, whether individual A-scans obtained in needle direction can contribute to the identification of pulmonary nodules.METHODS: OCT A-scans from freshly resected human lung tissue specimen were recorded through a customized needle with an embedded optical fiber. Bidirectional Long Short Term Memory networks (BLSTMs) were trained on randomly distributed training and test sets of the acquired A-scans. Patient specific training and different pre-processing steps were evaluated.RESULTS: Classification rates from 67.5% up to 76% were archived for different training scenarios. Sensitivity and specificity were highest for a patient specific training with 0.87 and 0.85. Low pass filtering decreased the accuracy from 73.2% on a reference distribution to 62.2% for higher cutoff frequencies and to 56% for lower cutoff frequencies.CONCLUSION: The results indicate that a grey value based classification is feasible and may provide additional information for diagnosis and navigation. Furthermore, the experiments show patient specific signal properties and indicate that the lower and upper parts of the frequency spectrum contribute to the classification.

AB - OBJECTIVES: Optical Coherence Tomography (OCT) has been proposed as a high resolution image modality to guide transbronchial biopsies. In this study we address the question, whether individual A-scans obtained in needle direction can contribute to the identification of pulmonary nodules.METHODS: OCT A-scans from freshly resected human lung tissue specimen were recorded through a customized needle with an embedded optical fiber. Bidirectional Long Short Term Memory networks (BLSTMs) were trained on randomly distributed training and test sets of the acquired A-scans. Patient specific training and different pre-processing steps were evaluated.RESULTS: Classification rates from 67.5% up to 76% were archived for different training scenarios. Sensitivity and specificity were highest for a patient specific training with 0.87 and 0.85. Low pass filtering decreased the accuracy from 73.2% on a reference distribution to 62.2% for higher cutoff frequencies and to 56% for lower cutoff frequencies.CONCLUSION: The results indicate that a grey value based classification is feasible and may provide additional information for diagnosis and navigation. Furthermore, the experiments show patient specific signal properties and indicate that the lower and upper parts of the frequency spectrum contribute to the classification.

KW - Biopsy, Needle

KW - Humans

KW - Image Interpretation, Computer-Assisted

KW - Image-Guided Biopsy

KW - Lung/pathology

KW - Multiple Pulmonary Nodules/classification

KW - Neural Networks, Computer

KW - Sensitivity and Specificity

KW - Software

KW - Tomography, Optical Coherence

U2 - 10.3414/ME13-01-0135

DO - 10.3414/ME13-01-0135

M3 - SCORING: Journal article

C2 - 24992968

VL - 53

SP - 245

EP - 249

JO - METHOD INFORM MED

JF - METHOD INFORM MED

SN - 0026-1270

IS - 4

ER -