MAFIA-CT: MAchine Learning Tool for Image Quality Assessment in Computed Tomography
- 152 Downloads
Different metrics are available for evaluating image quality (IQ) in computed tomography (CT). One of those is human observer studies, unfortunately they are time consuming and susceptible to variability. With these in mind, we developed a platform, based on deep learning, to optimise the work-flow and score IQ based human observations of low contrast lesions.
1476 images (from 43 CT devices) were used. The platform was evaluated for its accuracy, reliability and performance in both held-out tests, synthetic data and designed measurements. Synthetic data to evaluate the model capabilities and performance regarding varying structures and background. Designed measurements to evaluate the model performance in characterising CT protocols and devices regarding protocol dose and reconstruction.
We obtained 99.7% success rate on inlays detection and over 96% accuracy for given observer. From the synthetic data experiments, we observed a correlation between the minimum visible contrast and the lesion size; lesion’s contrast and visibility degradation due to noise levels; and no influence from external lesions to the central lesions detectability by the model. From the measurements in relation to dose, only between 20 and 25 mGy protocols differences were not statistically significant (p-values 0.076 and 0.408, respectively for 5 and 8 mm lesions). Additionally, our model showed improvements in IQ by using iterative reconstruction and the effect of reconstruction kernel.
Our platform enables the evaluation of large data-sets without the variability and time-cost associated with human scoring and subsequently providing a reliable and relatable metric for dose harmonisation and imaging optimisation in CT.
KeywordsComputed tomography Deep learning Image quality
We thank Prof. S. Scheidegger, Mr. C. Sommer, Mr. M. Weyland and Ms. C. Durán from the Zurich University of Applied Sciences, ZHAW (Winterthur, Switzerland) for the enlightening discussions, comprehensive support and for the phantom development. Additionally, we thank Mr. Michael Barnard for revising our work to improve the grammar and readability.
- Brankov, J.G., Wei, L., Yang, Y., Wernick, M.N.: Generalization evaluation of numerical observers for image quality assessment. In: 2006 IEEE Nuclear Science Symposium Conference Record, vol. 3, pp. 1696–1698 (2006). https://doi.org/10.1109/NSSMIC.2006.354225
- Le Coultre, R., et al.: Exposure of the swiss population by radiodiagnostics: 2013 review. Radiat. Prot. Dosimetry 169(1–4), 221–224 (2016)Google Scholar
- Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: a large-scale hierarchical image database. In: 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp. 248–255. IEEE (2009)Google Scholar
- Dibra, E., Melchior, S., Balkis, A., Wolf, T., Oztireli, C., Gross, M.: Monocular rgb hand pose inference from unsupervised refinable nets. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp. 1075–1085 (2018)Google Scholar
- Esteva, A., et al.: Dermatologist-level classification of skin cancer with deep neural networks. Nature 542, 115–119 (2017). https://doi.org/10.1038/nature21056
- Fischler, M.A., Bolles, R.C.: Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Commun. ACM 24(6), 381–395 (1981)Google Scholar
- Geyer, L.L., et al.: State of the art: iterative CT reconstruction techniques. Radiology 276(2), 339–357 (2015). https://doi.org/10.1148/radiol.2015132766
- Goenka, A.H., et al.: Image noise, cnr, and detectability of low-contrast, low-attenuation liver lesions in a phantom: effects of radiation exposure, phantom size, integrated circuit detector, and iterative reconstruction. Radiology 280(2), 475–482 (2016)Google Scholar
- Guimarães, L.S., et al.: Appropriate patient selection at abdominal dual-energy ct using 80 kv: relationship between patient size, image noise, and image quality. Radiology 257, 732–742 (2009)Google Scholar
- Hansis, E., Lorenz, C.: Landmark constellation models for medical image content identification and localization. Int. J. Comput. Assist. Radiol. Surg, 1–11 (2015). https://doi.org/10.1007/s11548-015-1328-5
- Iandola, F.N, et al.: Alexnet-level accuracy with 50x fewer parameters and \(<\)0.5 mb model size (2016). arXiv preprint arXiv:1602.07360
- Kopp, F.K., et al.: Cnn as model observer in a liver lesion detection task for x-ray computed tomography: a phantom study. Med. Phys. 45(10), 4439–4447 (2018)Google Scholar
- Kumar, A., Chellappa, R.: Disentangling 3D pose in a dendritic CNN for unconstrained 2D face alignment. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 430–439 (2018)Google Scholar
- Kyono, T., Gilbert, F.J., van der Schaar, M.: Mammo: a deep learning solution for facilitating radiologist-machine collaboration in breast cancer diagnosis (2018). ArXiv, abs/1811.02661Google Scholar
- Lima, T.V.L., Schindera, S., Scheidegger, S., Lutters, G.: Connecting the missing piece: a retrospective evaluation of image quality and dose in respect to the parameters variability for a clinical CT protocol. In: ECR 2018 Eurosafe Imaging (2018). https://doi.org/10.1594/esi2018/ESI-0071
- Liu, C.C., Qi, J.: Higher SNR PET image prediction using a deep learning model and MRI image. Phys. Med. Biol. 64(11), 115004 (2019)Google Scholar
- Liu, K., et al.: A gentle introduction to deep learning in medical image processing. Radiol. Artif. Intell. 1(3), 1–8 (2019)Google Scholar
- Maier, A., Syben, C., Lasser, T., Riess, C.: A gentle introduction to deep learning in medical image processing. Z Med. Phys. 29, 86–101 (2019)Google Scholar
- Rotzinger, D.C., et al.: Task-based model observer assessment of a partial model-based iterative reconstruction algorithm in thoracic oncologic multidetector CT. Sci. Rep. 8(17734) (2018)Google Scholar
- Samei, E., et al.: Medical imaging dose optimisation from ground up: expert opinion of an international summit. J. Radiol. Prot. 38, 967–989 (2018)Google Scholar
- Singh, S., et al.: Dose reduction and compliance with pediatric CT protocols adapted to patient size, clinical indication, and number of prior studies. Radiology 252, 200–208 (2009)Google Scholar
- Sommer, C., Icken, N., Özden, I., Lutters, G., Scheidegger, S.: Evaluation of low contrast resolution and radiation dose in abdominal CT protocols by a difference detail curve (DDC) method. Curr. Direct. Biomed. Eng. 3(2), 517–519 (2017)Google Scholar
- Verdun, F.R., et al.: Image quality in CT: from physical measurements to model observers. Physica Medica 31(8), 823–843 (2015)Google Scholar
- Wang, Y., et al.: Iterative quality enhancement via residual-artifact learning networks for low-dose CT. Phys. Med. Biol. 63(21), 215004 (2018)Google Scholar
- Yu, L., et al.: Prediction of human observer performance in a 2-alternative forced choice low-contrast detection task using channelized hotelling observer: impact of radiation dose and reconstruction algorithms. Medical Physics 40(4), 475–482 (2013)Google Scholar