It has been shown that the difference of squares cost function used by standard realignment packages (SPM and AIR) can lead to the detection of spurious activations, because the motion parameter estimations are biased by the activated areas. Therefore, this paper describes several experiments aiming at selecting a better similarity measure to drive functional magnetic resonance image registration. The behaviors of the Geman-McClure (GM) estimator, of the correlation ratio, and of the mutual information (MI) relative to activated areas are studied using simulated time series and actual data stemming from a 3T magnet. It is shown that these methods are more robust than the usual difference of squares measure. The results suggest also that the measures built from robust metrics like the GM estimator may be the best choice, while MI is also an interesting solution. Some more work, however, is required to compare the various robust metrics proposed in the literature.
We propose a 3D-2D image registration method that relates image features of 2D projection images to the transformation parameters of the 3D image by nonlinear regression. The method is compared with a conventional registration method based on iterative optimization. For evaluation, simulated X-ray images (DRRs) were generated from coronary artery tree models derived from 3D CTA scans. Registration of nine vessel trees was performed, and the alignment quality was measured by the mean target registration error (mTRE). The regression approach was shown to be slightly less accurate, but much more robust than the method based on an iterative optimization approach.
We evaluate the integration of 3D preoperative computed tomography angiography of the coronary arteries with intraoperative 2D X-ray angiographies by a recently proposed novel registration-by-regression method. The method relates image features of 2D projection images to the transformation parameters of the 3D image. We compared different sets of features and studied the influence of preprocessing the training set. For the registration evaluation, a gold standard was developed from eight X-ray angiography sequences from six different patients. The alignment quality was measured using the 3D mean target registration error (mTRE). The registration-by-regression method achieved moderate accuracy (median mTRE of 15 mm) on real images. It does therefore not provide yet a complete solution to the 3D-2D registration problem but it could be used as an initialisation method to eliminate the need for manual initialisation.
Convolutional neural networks (CNNs) have recently been successfully used in the medical field to detect and classify pathologies in different imaging modalities, including in mammography. One disadvantage of CNNs is the need for large training datasets, which are particularly difficult to obtain in the medical domain. One way to solve this problem is using a transfer learning approach, in which a CNN, previously pre-trained with a large amount of labelled non-medical data, is subsequently finetuned using a smaller dataset of medical data. In this paper, we use such a transfer learning approach, which is applied to three different networks that were pre-trained using the Imagenet dataset. We investigate how the performance of these pre-trained CNNs to classify lesions in mammograms is affected by the use, or not, of normalised images during the fine-tuning stage. We also assess the performance of a support vector machine fed with features extracted from the CNN and the combined use of handcrafted features to complement the CNN-extracted features. The obtained results are encouraging.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.