A Methodology to Automatically Segment 3D Ultrasonic Data Using X-ray Computed Tomography and a Convolutional Neural Network
Ultrasonic non-destructive testing (UT) is a proficient method for detecting damage in composite materials; however, conventional manual testing procedures are time-consuming and labor-intensive. We propose a semi-automated defect segmentation methodology employing a convolutional neural network (CN...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-05-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/13/10/5933 |
| _version_ | 1827742242890055680 |
|---|---|
| author | Juan-Ignacio Caballero Guillermo Cosarinsky Jorge Camacho Ernestina Menasalvas Consuelo Gonzalo-Martin Federico Sket |
| author_facet | Juan-Ignacio Caballero Guillermo Cosarinsky Jorge Camacho Ernestina Menasalvas Consuelo Gonzalo-Martin Federico Sket |
| author_sort | Juan-Ignacio Caballero |
| collection | DOAJ |
| description | Ultrasonic non-destructive testing (UT) is a proficient method for detecting damage in composite materials; however, conventional manual testing procedures are time-consuming and labor-intensive. We propose a semi-automated defect segmentation methodology employing a convolutional neural network (CNN) on 3D ultrasonic data, facilitated by the fusion of X-ray computed tomography (XCT) and Phased-Array Ultrasonic Testing (PAUT) data. This approach offers the ability to develop supervised datasets for cases where UT techniques inadequately assess defects and enables the creation of models with genuine defects rather than artificially introduced ones. During the training process, we recommend processing the 3D volumes as a sequence of 2D slices derived from each technique. Our methodology was applied to segment porosity, a common defect in composite materials, for which characteristics such as void size and shape remain immeasurable via UT. Precision, recall, F1 score, and Intersection over Union (IoU) metrics were used in the evaluation. The results of the evaluation show that the following challenges have to be faced for improvement: (i) achieving accurate 3D registration, (ii) discovering suitable similar keypoints for XCT and UT data registration, (iii) differentiating ultrasonic echoes originating from porosity versus those related to noise or microstructural features (interfaces, resin pockets, fibers, etc.), and, (iv) single out defect echoes located near the edges of the component. In fact, an average F1 score of 0.66 and IoU of 0.5 were obtained. |
| first_indexed | 2024-03-11T03:58:16Z |
| format | Article |
| id | doaj.art-077dad698b0b44a786f52645f3d5b7ba |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-11T03:58:16Z |
| publishDate | 2023-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-077dad698b0b44a786f52645f3d5b7ba2023-11-18T00:17:54ZengMDPI AGApplied Sciences2076-34172023-05-011310593310.3390/app13105933A Methodology to Automatically Segment 3D Ultrasonic Data Using X-ray Computed Tomography and a Convolutional Neural NetworkJuan-Ignacio Caballero0Guillermo Cosarinsky1Jorge Camacho2Ernestina Menasalvas3Consuelo Gonzalo-Martin4Federico Sket5Department of Materials Science, E. T. S. de Ingenieros de Caminos, Universidad Politecnica de Madrid (UPM), 28040 Madrid, SpainInstitute for Physical and Information Technologies (ITEFI), Spanish National Research Council (CSIC), c/Serrano 144, 28006 Madrid, SpainInstitute for Physical and Information Technologies (ITEFI), Spanish National Research Council (CSIC), c/Serrano 144, 28006 Madrid, SpainCentro de Tecnologia Biomedica, Universidad Politecnica de Madrid (UPM) Parque Científico y Tecnológico de la UPM, 28223 Madrid, SpainCentro de Tecnologia Biomedica, Universidad Politecnica de Madrid (UPM) Parque Científico y Tecnológico de la UPM, 28223 Madrid, SpainIMDEA Materiales c/ Eric Kandel 1, 28906 Getafe, SpainUltrasonic non-destructive testing (UT) is a proficient method for detecting damage in composite materials; however, conventional manual testing procedures are time-consuming and labor-intensive. We propose a semi-automated defect segmentation methodology employing a convolutional neural network (CNN) on 3D ultrasonic data, facilitated by the fusion of X-ray computed tomography (XCT) and Phased-Array Ultrasonic Testing (PAUT) data. This approach offers the ability to develop supervised datasets for cases where UT techniques inadequately assess defects and enables the creation of models with genuine defects rather than artificially introduced ones. During the training process, we recommend processing the 3D volumes as a sequence of 2D slices derived from each technique. Our methodology was applied to segment porosity, a common defect in composite materials, for which characteristics such as void size and shape remain immeasurable via UT. Precision, recall, F1 score, and Intersection over Union (IoU) metrics were used in the evaluation. The results of the evaluation show that the following challenges have to be faced for improvement: (i) achieving accurate 3D registration, (ii) discovering suitable similar keypoints for XCT and UT data registration, (iii) differentiating ultrasonic echoes originating from porosity versus those related to noise or microstructural features (interfaces, resin pockets, fibers, etc.), and, (iv) single out defect echoes located near the edges of the component. In fact, an average F1 score of 0.66 and IoU of 0.5 were obtained.https://www.mdpi.com/2076-3417/13/10/5933machine learningdeep learningultrasoundsphased arrayXCTdata fusion |
| spellingShingle | Juan-Ignacio Caballero Guillermo Cosarinsky Jorge Camacho Ernestina Menasalvas Consuelo Gonzalo-Martin Federico Sket A Methodology to Automatically Segment 3D Ultrasonic Data Using X-ray Computed Tomography and a Convolutional Neural Network Applied Sciences machine learning deep learning ultrasounds phased array XCT data fusion |
| title | A Methodology to Automatically Segment 3D Ultrasonic Data Using X-ray Computed Tomography and a Convolutional Neural Network |
| title_full | A Methodology to Automatically Segment 3D Ultrasonic Data Using X-ray Computed Tomography and a Convolutional Neural Network |
| title_fullStr | A Methodology to Automatically Segment 3D Ultrasonic Data Using X-ray Computed Tomography and a Convolutional Neural Network |
| title_full_unstemmed | A Methodology to Automatically Segment 3D Ultrasonic Data Using X-ray Computed Tomography and a Convolutional Neural Network |
| title_short | A Methodology to Automatically Segment 3D Ultrasonic Data Using X-ray Computed Tomography and a Convolutional Neural Network |
| title_sort | methodology to automatically segment 3d ultrasonic data using x ray computed tomography and a convolutional neural network |
| topic | machine learning deep learning ultrasounds phased array XCT data fusion |
| url | https://www.mdpi.com/2076-3417/13/10/5933 |
| work_keys_str_mv | AT juanignaciocaballero amethodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT guillermocosarinsky amethodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT jorgecamacho amethodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT ernestinamenasalvas amethodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT consuelogonzalomartin amethodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT federicosket amethodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT juanignaciocaballero methodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT guillermocosarinsky methodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT jorgecamacho methodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT ernestinamenasalvas methodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT consuelogonzalomartin methodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork AT federicosket methodologytoautomaticallysegment3dultrasonicdatausingxraycomputedtomographyandaconvolutionalneuralnetwork |