New Model of Heteroasociative min Memory Robust to Acquisition Noise

Associative memories in min and max algebra are of great interest for pattern recognition. One property of these is that they are one-shot, that is, in an attempt they converge to the solution without having to iterate. These memories have proven to be very efficient, but they manifest some weakness with mixed noise. If an appropriate kernel is not used, that is, a subset of the pattern to be recalled that is not affected by noise, memories fail noticeably. A possible problem for building kernels with sufficient conditions, using binary and gray-scale images, is not knowing how the noise is registered in these images. A solution to this problem is presented by analyzing the behavior of the acquisition noise. What is new about this analysis is that, noise can be mapped to a distance obtained by a distance transform. Furthermore, this analysis provides the basis for a new model of min heteroassociative memory that is robust to the acquisition/mixed noise. The proposed model is novel because min associative memories are typically inoperative to mixed noise. The new model of heteroassocitative memory obtains very interesting results with this type of noise.

Distance transform-watershed segmentation and multi-layer perceptron algorithm to separate touching orange fruit in digital images

An algorithm to separate touching oranges using a distance transform-watershed segmentation is presented. In this study, there are four classes of oranges, such as class A, B, C, and D, respectively. The size of each class is based on the Indonesian National Standard (SNI), the sample used is 168 oranges of which 140 are for training and 28 oranges are for testing. The image of citrus fruits was captured using Kinect v2 camera with a camera resolution of 1920 × 1080 pixels, the distance from the camera to the background is 23 cm. The images were captured in PNG format. The watersheds were computed based on the distance transformed by orange regions. The corresponding basins were finally used to split the falsely connected corn kernel by intersecting the basins with the corn kernel regions. Experimental results show that the multi-layer perceptrons have classification accuracy rates of 92.85%. The algorithm appears to be robust enough to separate most of the multiple touching scenarios.

Deteksi Gulma Berdasarkan Warna HSV dan Fitur Bentuk Menggunakan Jaringan Syaraf Tiruan

<p class="Abstrak">Gulma merupakan tanaman pengganggu dalam lahan pertanian. Herbisida merupakan obat yang efektif membunuh gulma tersebut. Penyemprotan herbisida harus tepat sasaran kepada gulma saja dan tidak mengenai tanaman. Penelitian ini membuat sistem yang dapat mendeteksi gulma secara otomatis di antara tanaman pada lahan pertanian riil. Sistem ini menggunakan gambar lahan pertanian riil dimana tanaman tampak utuh (daun dapat lebih dari satu) yang diambil menggunakan kamera dengan posisi vertikal menghadap ke bawah. Algoritma yang dibuat menggunakan segmentasi berdasarkan warna hijau dalam ruang warna HSV untuk mendeteksi daun, baik gulma maupun tanaman pada beragam pencahayaan. Sebanyak tiga fitur bentuk domain spasial digunakan untuk membedakan gulma dengan tanaman yang memiliki karakteristik bentuk daun yang berbeda. Fitur bentuk yang digunakan adalah <em>Rectangularity, Edge-to-Center distances function</em>, dan <em>Distance Transform function</em>. Klasifikasi gulma dan tanaman menggunakan metode Jaringan syaraf tiruan (JST) yang dapat dilatih secara <em>offline. </em>Dari 149 tanaman yang terdeteksi dimana 70% sebagai data training, 15% data validasi dan 15% data uji, didapati akurasi pengujian sebesar 95.46%.</p><p class="Abstrak"><em><strong><br /></strong></em></p><p class="Abstrak"><em><strong>Abstract</strong></em></p><p class="Abstract"><em>Weed is a major challenge in a crop plantation. A herbicide is the most effective substance to kill this unwanted vegetation. Spraying the herbicide must be done carefully to target the weeds only. Here in this research, we develop an algorithm that detects weeds among the plants based on the shape of their leaves. The detection is based on images that were acquired using a camera. The leaves of weeds and plants were detected based on their green color using segmentation in HSV color-space as it is more effective to detect objects in various illumination. Three shape features were extracted, which are Rectangularity that is based on Rectangularity, Edge-to-Center distance function, and Distance Transform function. Those features were fed into a learning algorithm, Artificial Neural Network (ANN), to classify whether it is the plant or the weed. The testing on the weed classification in a real outdoor environment showed 95.46% accuracy using a total of 149 detected plants (70% as training data, 15% as validation data, and 15% as testing data).<strong></strong></em></p><p class="Abstrak"><em><strong><br /></strong></em></p>

Automatic Segmentation of Specific Intervertebral Discs through a Two-Stage MultiResUNet Model

The automatic segmentation of intervertebral discs from medical images is an important task for an intelligent clinical system. In this study, a deep learning model based on the MultiResUNet model for the automatic segmentation of specific intervertebral discs is presented. MultiResUNet can easily segment all intervertebral discs in MRI images; however, when only certain specific intervertebral discs need to be segmented, problems with segmentation errors, misalignment, and noise occur. In order to solve these problems, a two-stage MultiResUNet model is proposed. Connected-component labeling, automatic cropping, and distance transform are used in the proposed method. The experimental results show that the segmentation errors and misalignments of specific intervertebral discs are greatly reduced, and the segmentation accuracy is increased to about 94%. The performance of the proposed method proves its usefulness for the automatic segmentation of specific intervertebral discs over other deep learning models, such as the U-Net, CNN-based, Attention U-Net, and MultiResUNet models.

Detection of 3D Arterial Centerline Extraction in Spiral CT Coronary Angiography

This paper presents an in-depth study and analysis of the 3D arterial centerline in spiral CT coronary angiography, and constructs its detection and extraction technique. The first time, the distance transform is used to complete the boundary search of the original figure; the second time, the distance transform is used to calculate the value of the distance transform of all voxels, and according to the value of the distance transform, unnecessary voxels are deleted, to complete the initial contraction of the vascular region and reduce the computational consumption in the next process; then, the nonwitnessed voxels are used to construct the maximum inner joint sphere model and find the skeletal voxels that can reflect the shape of the original figure. Finally, the skeletal lines were optimized on these initially extracted skeletal voxels using a dichotomous-like principle to obtain the final coronary artery centerline. Through the evaluation of the experimental results, the algorithm can extract the coronary centerline more accurately. In this paper, the segmentation method is evaluated on the test set data by two kinds of indexes: one is the index of segmentation result evaluation, including dice coefficient, accuracy, specificity, and sensitivity; the other is the index of clinical diagnosis result evaluation, which is to refine the segmentation result for vessel diameter detection. The results obtained in this paper were compared with the physicians’ labeling results. In terms of network performance, the Dice coefficient obtained in this paper was 0.89, the accuracy was 98.36%, the sensitivity was 93.36%, and the specificity was 98.76%, which reflected certain advantages in comparison with the advanced methods proposed by previous authors. In terms of clinical evaluation indexes, by performing skeleton line extraction and diameter calculation on the results obtained by the segmentation method proposed in this paper, the absolute error obtained after comparing with the diameter of the labeled image was 0.382 and the relative error was 0.112, which indicates that the segmentation method in this paper can recover the vessel contour more accurately. Then, the results of coronary artery centerline extraction with and without fine branch elimination were evaluated, which proved that the coronary artery centerline has higher accuracy after fine branch elimination. The algorithm is also used to extract the centerline of the complete coronary artery tree, and the results prove that the algorithm has better results for the centerline extraction of the complete coronary vascular tree.

Impact of Dental Model Height on Thermoformed PET-G Aligner Thickness—An In Vitro Micro-CT Study

The aim of the present study was to investigate whether base height of 3D-printed dental models has an impact on local thickness values from polyethylene terephthalate glycol (PET-G) aligners. A total of 20 aligners were thermoformed on dental models from the upper jaw exhibiting either a 5 mm high (H) or narrow (N), i.e., 0 mm, base height. The aligners were digitized using micro-CT, segmented, and local thickness values were computed utilizing a 3D-distance transform. The mean thickness values and standard deviations were assessed for both groups, and local thickness values at pre-defined reference points were also recorded. The statistical analysis was performed using R. Aligners in group H were significantly thinner and more homogenous compared to group N (p < 0.001). Significant differences in thickness values were observed among tooth types between both groups. Whereas thickness values were comparable at cusp tips and occlusal/incisal/cervical measurement locations, facial and palatal surfaces were significantly thicker in group N compared to group H (p < 0.01). Within the limits of the study, the base height of 3D-printed models impacts on local thickness values of thermoformed aligners. The clinician should consider potential implication on exerted forces at the different tooth types, and at facial as well as palatal surfaces.

WDTISeg: One-Stage Interactive Segmentation for Breast Ultrasound Image Using Weighted Distance Transform and Shape-Aware Compound Loss

Accurate tumor segmentation is important for aided diagnosis using breast ultrasound. Interactive segmentation methods can obtain highly accurate results by continuously optimizing the segmentation result via user interactions. However, traditional interactive segmentation methods usually require a large number of interactions to make the result meet the requirements due to the performance limitations of the underlying model. With greater ability in extracting image information, convolutional neural network (CNN)-based interactive segmentation methods have been shown to effectively reduce the number of user interactions. In this paper, we proposed a one-stage interactive segmentation framework (interactive segmentation using weighted distance transform, WDTISeg) for breast ultrasound image using weighted distance transform and shape-aware compound loss. First, we used a pre-trained CNN to attain an initial automatic segmentation, based on which the user provided interaction points of mis-segmented areas. Then, we combined Euclidean distance transform and geodesic distance transform to convert interaction points into weighted distance maps to transfer segmentation guidance information to the model. The same CNN accepted the input image, the initial segmentation, and weighted distance maps as a concatenation input and provided a refined result, without another additional segmentation network. In addition, a shape-aware compound loss function using prior knowledge was designed to reduce the number of user interactions. In the testing phase on 200 cases, our method achieved a dice of 82.86 ± 16.22 (%) for automatic segmentation task and a dice of 94.45 ± 3.26 (%) for interactive segmentation task after 8 interactions. The results of comparative experiments proved that our method could obtain higher accuracy with fewer simple interactions than other interactive segmentation methods.