Cascaded multi-scale convolutional encoder-decoders for breast mass segmentation in high-resolution mammograms.

This paper addresses breast mass segmentation from high-resolution mammograms. To cope with strong class imbalance, huge diversity of size, shape, texture and contour as well as limited receptive field, mass segmentation is achieved through a multi-scale cascade of deep convolutional encoder-decoders without any pre-detection scheme. Multi-scale information is integrated using auto-context to make long-range spatial context arising from lower scale impact training at higher resolution. The pipeline is trained end-to-end to benefit from simultaneous segmentation refinement performed at each level. It incorporates transfer learning and fine tuning from DDSM-CBIS to INbreast datasets to further improve mass delineations. The comprehensive evaluation provided for high-resolution INbreast images highlights promising model generalizability against standard encoder-decoder strategies.

[Assessment of a 3D digital orthoptic test platform]. / Évaluation d'une plateforme d'orthoptie numérique 3D.

PURPOSE: To compare standard orthoptic tests with a novel digital 3D orthoptic platform, 3DeltaEasy(©) from Orthoptica(®). MATERIALS AND METHODS: This study tests the 3D digital orthoptics platform, 3DeltaEasy(©) from Orthoptica(®) and compares it to the corresponding standard orthoptic tests. This platform consists of a computer equipped with dedicated software, a video projector and 3D liquid crystal glasses. Three tests were compared: Wirt test, measurement of horizontal and vertical phorias, and the horizontal fusional amplitude in convergence and divergence. A total of 102 subjects, 53 males (52 %) and 49 females (48 %), aged between 9 years and 72 years (mean age 33±16.4 years) were examined at the ophthalmologic department of the Brest Hospital (France) and included in this observational cross-sectional study. Subjects recruited in this study were patients requiring orthoptic screening or therapy. Patients without their optimal visual corrections were excluded. All patients underwent both ophthalmological and orthoptic examination including Wirt fly stereotest with polarizing spectacles, cover tests to evaluate and measure the horizontal and vertical deviation of the lines of sight, horizontal vergence ranges using prism bar and their equivalent tests implemented in the digital 3D orthoptic tests 3DeltaEasy(©) from Orthoptica(®). RESULTS: All data were processed using MedCalc Statistical Software version 14.12.0 (MedCalc Software bvba, Ostend, Belgium). The main result of this study is that 3DeltaEasy(©) and the classical Wirt test are correlated (Spearman's coefficient of rank correlation: ρ=0.74; P<0.0001), cover tests are equivalent for intermediate and far vision (paired t-test; P=0.46 and P=0.51), and horizontal and vertical vergence range are comparable for distance vision (paired t-test; P=0.34 and P=0.94). CONCLUSION: New digital 3D tools could easily substitute for some orthoptic tests with better ergonomics. Eventually, by increasing the number of tests performed, it could substitute for nearly all tests.

Automatic correction of rotating ultrasound bio microscopy acquisitions for the segmentation of the eye anterior segment.

We have developed a rotating 3D probe prototype in order to acquire the anterior segment of the eye in three dimensions. The acquisition accuracy has to be sufficient to allow for the use of automatic segmentation of the provided data, and thus generate a 3D structure of the eye, for which it could be easier to obtain measurements than in 2D images. We have created an image post processing scheme in order to compensate for vibrations and eye movements during acquisition that are associated with increased noise. These tools have been applied to 92 volume datasets acquired on 21 patients in pre-operative conditions. Acquisition noise was reduced by 97% in specific conditions with respect to data acquired without correction.

Wavelet optimization for content-based image retrieval in medical databases.

We propose in this article a content-based image retrieval (CBIR) method for diagnosis aid in medical fields. In the proposed system, images are indexed in a generic fashion, without extracting domain-specific features: a signature is built for each image from its wavelet transform. These image signatures characterize the distribution of wavelet coefficients in each subband of the decomposition. A distance measure is then defined to compare two image signatures and thus retrieve the most similar images in a database when a query image is submitted by a physician. To retrieve relevant images from a medical database, the signatures and the distance measure must be related to the medical interpretation of images. As a consequence, we introduce several degrees of freedom in the system so that it can be tuned to any pathology and image modality. In particular, we propose to adapt the wavelet basis, within the lifting scheme framework, and to use a custom decomposition scheme. Weights are also introduced between subbands. All these parameters are tuned by an optimization procedure, using the medical grading of each image in the database to define a performance measure. The system is assessed on two medical image databases: one for diabetic retinopathy follow up and one for screening mammography, as well as a general purpose database. Results are promising: a mean precision of 56.50%, 70.91% and 96.10% is achieved for these three databases, when five images are returned by the system.

[Peribulbar anesthesia: efficacy of a single injection with a limited local anesthetic volume]. / Anesthésie péribulbaire: efficacité d'une seule injection et d'un volume d'anesthésiques locaux limité.

INTRODUCTION: Cataract surgery can be performed with peribulbar anesthesia. The classical technique consists of two injections of local anesthetics. The purpose of our study was to assess peribulbar anesthesia with a single injection and a limited volume of local anesthetics. MATERIAL AND METHOD: After local ethics committee agreement and oral consent, patients scheduled for cataract surgery using peribulbar anesthesia were prospectively included. The lower temporal puncture was performed with a peribulbar needle with propofol sedation. The mixture of local anesthetics was administered with tactile control of orbital pressure. The puncture was followed by a 10-min compression of the ocular globe. Akinesia, analgesia, complications, and surgical conditions were noted. RESULTS: A total of 101 successive patients were included. We administered 1.2 mg/kg of propofol. The volume of local anesthetics administered was 5.0 +/- 0.9 ml. Ninety patients had akinesia at 10 min and 6.7% moderate chemosis. No puncture complication occurred. At the end of surgery, the pain noted by the patients was 0.4 +/- 2.1 out of 100 (range, 0-10). Surgical conditions were good for all patients. CONCLUSION: Peribulbar anesthesia performed with a single injection and a limited volume of local anesthetics allows cataract surgery in good conditions for the surgeon with very good analgesia for the patient.

A low distorsion and reversible watermark: application to angiographic images of the retina.

Medical image security can be enhanced using watermarking, which allows embedding the protection information as a digital signature, by modifying the pixel gray levels of the image. In this paper we propose a reversible watermarking scheme which guarantees that once the embedded message is read, alterations introduced during the insertion process can be removed from the image. Thereafter, original pixel gray levels of the image are restored. The proposed approach relies on estimation of image signal that is invariant to the insertion process, and permits to introduce a very slight watermark within the image. In fact, the insertion process adds or subtracts at least one gray level to the pixels of the original image. Depending on the image to be watermarked, in our case angiographic images of the retina, it is expected that such image alteration will not have any impact on the diagnosis quality, and consequently that the watermark can be kept within the image while this one is interpreted.

[Modeling the eye based on simulated refractive surgery]. / Modélisation de l'oeil en vue de simulations de chirurgies réfractives.

PURPOSE: To achieve three-dimensional modelizing of the eyeball (morphological and mechanical behavior) in order to simulate the impact of various refractive surgery techniques and to study the normal and pathological states of the eye. METHOD: Rebuilding the ocular shell is based on different kinds of imaging (MRI, ultrasound) including information provided by video topography. Image data are treated using suitable numerized filters that allow automatic segmentations of ocular globus edges. Reconstruction is based on specific mathematical functions (B-splines). The mechanical behavior of a reconstructed model is simulated by solving equations of linearized elasticity with the finitude elements method. RESULTS: Numerous simulations mimmed different refractive surgical techniques and, then validated the model. In addition, simulations of various pathologies allowed us to verify certain clinical hypotheses. CONCLUSION: This work, although still experimental, demonstrates the advantages of such simulations and will allow novice physicians an easier approach to different surgical techniques and will help them understand their effect. Furthermore, it might be useful for simulation of new surgical concepts even before their in vivo evaluation.