This paper presents a method for fast computation of Hessian-based enhancement filters, whose conditions for identifying particular structures in medical images are associated only with the signs of Hessian eigenvalues. The computational costs of Hessian-based enhancement filters come mainly from the computation of Hessian eigenvalues corresponding to image elements to obtain filter responses, because computing eigenvalues of a matrix requires substantial computational effort. High computational cost has become a challenge in the application of Hessian-based enhancement filters. Using a property of the characteristic polynomial coefficients of a matrix and the well-known Routh-Hurwitz criterion in control engineering, it is shown that under certain conditions, the response of a Hessian-based enhancement filter to an image element can be obtained without having to compute Hessian eigenvalues. The computational cost can thus be reduced. Experimental results on several medical images show that the method proposed in this paper can reduce significantly the number of computations of Hessian eigenvalues and the processing times of images. The percentage reductions of the number of computations of Hessian eigenvalues for enhancing blob- and tubular-like structures in two-dimensional images are approximately 90% and 65%, respectively. For enhancing blob-, tubular-, and plane-like structures in three-dimensional images, the reductions are approximately 97%, 75%, and 12%, respectively. For the processing times, the percentage reductions for enhancing blob- and tubular-like structures in two-dimensional images are approximately 31% and 7.5%, respectively. The reductions for enhancing blob-, tubular-, and plane-like structures in three-dimensional images are approximately 68%, 55%, and 3%, respectively.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1016/j.cmpb.2014.05.002 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Effective feature selection based HOBS pruned- ELM model for tomato plant leaf disease classification.
PLoS One
December 2024
School of Computer Science Engineering and Information Systems, Vellore Institute of Technology, Vellore, Tamil Nādu, India.
Tomato cultivation is expanding rapidly, but the tomato sector faces significant challenges from various sources, including environmental (abiotic stress) and biological (biotic stress or disease) threats, which adversely impact the crop's growth, reproduction, and overall yield potential. The objective of this work is to build deep learning based lightweight convolutional neural network (CNN) architecture for the real-time classification of biotic stress in tomato plant leaves. This model proposes to address the drawbacks of conventional CNNs, which are resource-intensive and time-consuming, by using optimization methods that reduce processing complexity and enhance classification accuracy.
View Article and Find Full Text PDFPSAA-nnUNet: An Efficient Method for CT Carotid Artery Image Segmentation.
Adv Exp Med Biol
October 2024
School of Information Science and Technology, Nantong University, Nantong, Jiangsu Province, China.
Article Synopsis
- Carotid artery stenosis (CAS) is a major risk factor for silent but serious ischaemic strokes, making early detection crucial for improving patient outcomes.
- This study introduces a non-invasive assessment method using deep learning and CT imaging to detect CAS early, consisting of three components: enhancing vascular structures, utilizing a novel neural network for segmentation, and evaluating CAS severity with a standard formula.
- The proposed PSAA-nnUNet algorithm demonstrated a segmentation accuracy of 0.82, highlighting its effectiveness and potential for real-world medical applications.
A Hessian-Based Technique for Specular Reflection Detection and Inpainting in Colonoscopy Images.
IEEE J Biomed Health Inform
August 2024
In the field of Computer-Aided Detection (CADx), the use of AI-based algorithms for disease detection in endoscopy images, especially colonoscopy images, is on the rise. However, these algorithms often encounter performance issues due to obstructions like specular reflection, resulting in false positives. This paper presents a novel algorithm specifically designed to tackle the challenges posed by high specular reflection regions in colonoscopy images.
View Article and Find Full Text PDFSemi-automated algorithm using directional filter for the precise quantification of non-perfusion area on widefield swept-source optical coherence tomography angiograms.
Quant Imaging Med Surg
June 2023
Retina Service, Massachusetts Eye and Ear, Department of Ophthalmology, Harvard Medical School, Boston, MA, USA.
Background: The clinical application of optical coherence tomography angiography (OCTA) has been well documented in literature with its promising potential in dye-less evaluation of various retinal vascular pathologies. Recent advances in OCTA help us gather wider field of view with 12 mm × 12 mm and montage compared to the standard dye-based scans, which has a higher accuracy and sensitivity in detection of peripheral pathologies. The aim of this study is to build a semi-automated algorithm to precisely quantify the non-perfusion areas (NPAs) on widefield swept-source optical coherence tomography angiography (WF SS-OCTA).
View Article and Find Full Text PDFArticle Synopsis
- The study focuses on improving the visualization of neovascularization in injured finger tendons using a novel high-frequency micro-Doppler imaging (HFμDI) technique instead of traditional ultrasound, which lacks the necessary resolution.
- The HFμDI method was tested on both small animal models and human patients, showing superior imaging capabilities and enhanced visibility of microvasculature in injuries.
- Results indicated significant neovascularization early in the healing process, which reduced after rehabilitation, suggesting that physical therapy can positively affect tendon healing by lowering neovascular density.
Want AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!