Displaying all 18 publications

  1. Adeshina AM, Hashim R, Khalid NE
    Interdiscip Sci, 2014 Sep;6(3):222-34.
    PMID: 25205500 DOI: 10.1007/s12539-013-0204-7
    Hepatocellular Carcinoma is the most common type of liver cancer having a strong relation with cirrhosis. Undoubtedly, cirrhosis may be caused by the virus infection of hepatitis B (HBV) and hepatitis C (HBC) or through alchoholism. However, even when cirrhosis has not been developed, patients with hepatitis viral infections are still at the risk of liver cancer. Apparently, among the numerous medical imaging techniques, Computed Tomography (CT) is the best in defining liver tumor borders. Unfortunately, these imaging techniques, including the CT procedures, usually rely on an appended application to reconstruct the generated 2-D slices to 3-D model. This may involve high performance computation, may be time-consuming or costly. Moreover, even with the outstanding performances of CT in defining the liver tumor boundaries, contrast between tumor tissues and the surrounding liver parenchyma is too low in CT slices. With such a close proxity in the tumor and the surrounding liver tissues, accurate characterization of liver tumor is a challenge. Previously, algorithms were developed to reveal abnormalities in brain's MRI datasets and CT abdominal pelvic, however, introducing a framework that could accurately characterize liver tumor and its surrounding tissues in CT datasets would go a long way in contributing to medical diagnosis and therapy planning of Hepatocellular Carcinoma. This paper proposes an Hepatocellular Carcinoma framework by extending the functionalities of SurLens Visualization System with an automatic liver tumor localization technique using Compute Unified Device Architecture (CUDA). The study was evaluated with liver CT datasets from the Imaging Science and Information Systems (ISIS) Center, the Georgetown University Medical Center. Significantly, visualization of liver CT datasets and the localization of the entangled tumor was achieved without prior datasets segmentation. Interestingly, the framework achieved remarkably good processing speed at a reasonably cheaper cost with an immediate reconstruction of the datasets and mapping of the tumor tissues within the surrounding liver parenchyma.
  2. Adeshina AM, Hashim R, Khalid NE, Abidin SZ
    Interdiscip Sci, 2013 Mar;5(1):23-36.
    PMID: 23605637 DOI: 10.1007/s12539-013-0155-z
    In the medical diagnosis and treatment planning, radiologists and surgeons rely heavily on the slices produced by medical imaging devices. Unfortunately, these image scanners could only present the 3-D human anatomical structure in 2-D. Traditionally, this requires medical professional concerned to study and analyze the 2-D images based on their expert experience. This is tedious, time consuming and prone to error; expecially when certain features are occluding the desired region of interest. Reconstruction procedures was earlier proposed to handle such situation. However, 3-D reconstruction system requires high performance computation and longer processing time. Integrating efficient reconstruction system into clinical procedures involves high resulting cost. Previously, brain's blood vessels reconstruction with MRA was achieved using SurLens Visualization System. However, adapting such system to other image modalities, applicable to the entire human anatomical structures, would be a meaningful contribution towards achieving a resourceful system for medical diagnosis and disease therapy. This paper attempts to adapt SurLens to possible visualisation of abnormalities in human anatomical structures using CT and MR images. The study was evaluated with brain MR images from the department of Surgery, University of North Carolina, United States and CT abdominal pelvic, from the Swedish National Infrastructure for Computing. The MR images contain around 109 datasets each of T1-FLASH, T2-Weighted, DTI and T1-MPRAGE. Significantly, visualization of human anatomical structure was achieved without prior segmentation. SurLens was adapted to visualize and display abnormalities, such as an indication of walderstrom's macroglobulinemia, stroke and penetrating brain injury in the human brain using Magentic Resonance (MR) images. Moreover, possible abnormalities in abdominal pelvic was also visualized using Computed Tomography (CT) slices. The study shows SurLens' functionality as a 3-D Multimodal Visualization System.
  3. Adeshina AM, Hashim R, Khalid NE, Abidin SZ
    Interdiscip Sci, 2012 Sep;4(3):161-72.
    PMID: 23292689 DOI: 10.1007/s12539-012-0132-y
    CT and MRI scans are widely used in medical diagnosis procedures, but they only produce 2-D images. However, the human anatomical structure, the abnormalities, tumors, tissues and organs are in 3-D. 2-D images from these devices are difficult to interpret because they only show cross-sectional views of the human structure. Consequently, such circumstances require doctors to use their expert experiences in the interpretation of the possible location, size or shape of the abnormalities, even for large datasets of enormous amount of slices. Previously, the concept of reconstructing 2-D images to 3-D was introduced. However, such reconstruction model requires high performance computation, may either be time-consuming or costly. Furthermore, detecting the internal features of human anatomical structure, such as the imaging of the blood vessels, is still an open topic in the computer-aided diagnosis of disorders and pathologies. This paper proposes a volume visualization framework using Compute Unified Device Architecture (CUDA), augmenting the widely proven ray casting technique in terms of superior qualities of images but with slow speed. Considering the rapid development of technology in the medical community, our framework is implemented on Microsoft.NET environment for easy interoperability with other emerging revolutionary tools. The framework was evaluated with brain datasets from the department of Surgery, University of North Carolina, United States, containing around 109 MRA datasets. Uniquely, at a reasonably cheaper cost, our framework achieves immediate reconstruction and obvious mappings of the internal features of human brain, reliable enough for instantaneous locations of possible blockages in the brain blood vessels.
  4. Kaur H, Ahmad M, Scaria V
    Interdiscip Sci, 2016 Mar;8(1):95-101.
    PMID: 26298582 DOI: 10.1007/s12539-015-0273-x
    There is emergence of multidrug-resistant Salmonella enterica serotype typhi in pandemic proportions throughout the world, and therefore, there is a necessity to speed up the discovery of novel molecules having different modes of action and also less influenced by the resistance formation that would be used as drug for the treatment of salmonellosis particularly typhoid fever. The PhoP regulon is well studied and has now been shown to be a critical regulator of number of gene expressions which are required for intracellular survival of S. enterica and pathophysiology of disease like typhoid. The evident roles of two-component PhoP-/PhoQ-regulated products in salmonella virulence have motivated attempts to target them therapeutically. Although the discovery process of biologically active compounds for the treatment of typhoid relies on hit-finding procedure, using high-throughput screening technology alone is very expensive, as well as time consuming when performed on large scales. With the recent advancement in combinatorial chemistry and contemporary technique for compounds synthesis, there are more and more compounds available which give ample growth of diverse compound library, but the time and endeavor required to screen these unfocused massive and diverse library have been slightly reduced in the past years. Hence, there is demand to improve the high-quality hits and success rate for high-throughput screening that required focused and biased compound library toward the particular target. Therefore, we still need an advantageous and expedient method to prioritize the molecules that will be utilized for biological screens, which saves time and is also inexpensive. In this concept, in silico methods like machine learning are widely applicable technique used to build computational model for high-throughput virtual screens to prioritize molecules for advance study. Furthermore, in computational analysis, we extended our study to identify the common enriched structural entities among the biologically active compound toward finding out the privileged scaffold.
  5. Ashwinder K, Kho MT, Chee PM, Lim WZ, Yap IK, Choi SB, et al.
    Interdiscip Sci, 2015 Aug 22.
    PMID: 26297309
    Heat shock proteins (Hsps) 60 and 70 are postulated as a potential drug target for toxoplasmosis due to its importance in the developmental and survival of Toxoplasma gondii (T. gondii). As of today, there have been no reports on three-dimensional (3D) structure of Hsp60 and Hsp70 deposited in the Brookhaven Protein Data Bank. Hence, this study was conducted to predict 3D structures for Hsp60 and Hsp70 in T. gondii by homology modeling. Selection of the best predicted model was done based on multiple scoring functions. In addition, virtual screening was performed to short-list chemical compounds from the National Cancer Institute (NCI) Diversity Set III in search of potential inhibitor against Hsp60 and Hsp70 in T. gondii. Prior to virtual screening, binding sites of Hsp60 and Hsp70 were predicted using various servers and were used as the center in docking studies. The Hsps were docked against known natural ligands to validate the method used in estimating free energy of binding (FEB) and possible interactions between ligand and protein. Virtual screening was performed with a total of 1560 compounds from the NCI Diversity Set III. The compounds were ranked subsequently according to their FEB. Molecular basis of interactions of the top five ranked compounds was investigated using Ligplot(+). The major interactions exhibited were hydrogen bonding and hydrophobic interactions in binding to Hsp60 and Hsp70. The results obtained provided information and guidelines for the development of inhibitors for Hsp60 and Hsp70 in T. gondii.
  6. Ismail NA, Jusoh SA
    Interdiscip Sci, 2017 Dec;9(4):499-511.
    PMID: 26969331 DOI: 10.1007/s12539-016-0157-8
    Dengue infections are currently estimated to be 390 million cases annually. Yet, there is no vaccine or specific therapy available. Envelope glycoprotein E (E protein) of DENV mediates viral attachment and entry into the host cells. Several flavonoids have been shown to inhibit HIV-1 and hepatitis C virus entry during the virus-host membrane fusion. In this work, molecular docking method was employed to predict the binding of nine flavonoids (baicalin, baicalein, EGCG, fisetin, glabranine, hyperoside, ladanein, quercetin and flavone) to the soluble ectodomain of DENV type 2 (DENV2) E protein. Interestingly, eight flavonoids were found to dock into the same binding pocket located between the domain I and domain II of different subunits of E protein. Consistent docking results were observed not only for the E protein structures of the DENV2-Thai and DENV2-Malaysia (a homology model) but also for the E protein structures of tick-borne encephalitis virus and Japanese encephalitis virus. In addition, molecular dynamics simulations were performed to further evaluate the interaction profile of the docked E protein-flavonoid complexes. Ile4, Gly5, Asp98, Gly100 and Val151 residues of the DENV2-My E protein that aligned to the same residues in the DENV2-Thai E protein form consistent hydrogen bond interactions with baicalein, quercetin and EGCG during the simulations. This study demonstrates flavonoids potentially form interactions with the E protein of DENV2.
  7. Adeshina AM, Hashim R
    Interdiscip Sci, 2017 Mar;9(1):140-152.
    PMID: 26754740 DOI: 10.1007/s12539-015-0140-9
    Diagnostic radiology is a core and integral part of modern medicine, paving ways for the primary care physicians in the disease diagnoses, treatments and therapy managements. Obviously, all recent standard healthcare procedures have immensely benefitted from the contemporary information technology revolutions, apparently revolutionizing those approaches to acquiring, storing and sharing of diagnostic data for efficient and timely diagnosis of diseases. Connected health network was introduced as an alternative to the ageing traditional concept in healthcare system, improving hospital-physician connectivity and clinical collaborations. Undoubtedly, the modern medicinal approach has drastically improved healthcare but at the expense of high computational cost and possible breach of diagnosis privacy. Consequently, a number of cryptographical techniques are recently being applied to clinical applications, but the challenges of not being able to successfully encrypt both the image and the textual data persist. Furthermore, processing time of encryption-decryption of medical datasets, within a considerable lower computational cost without jeopardizing the required security strength of the encryption algorithm, still remains as an outstanding issue. This study proposes a secured radiology-diagnostic data framework for connected health network using high-performance GPU-accelerated Advanced Encryption Standard. The study was evaluated with radiology image datasets consisting of brain MR and CT datasets obtained from the department of Surgery, University of North Carolina, USA, and the Swedish National Infrastructure for Computing. Sample patients' notes from the University of North Carolina, School of medicine at Chapel Hill were also used to evaluate the framework for its strength in encrypting-decrypting textual data in the form of medical report. Significantly, the framework is not only able to accurately encrypt and decrypt medical image datasets, but it also successfully encrypts and decrypts textual data in Microsoft Word document, Microsoft Excel and Portable Document Formats which are the conventional format of documenting medical records. Interestingly, the entire encryption and decryption procedures were achieved at a lower computational cost using regular hardware and software resources without compromising neither the quality of the decrypted data nor the security level of the algorithms.
  8. Al-Ameen Z, Sulong G
    Interdiscip Sci, 2015 Sep;7(3):319-25.
    PMID: 26199211 DOI: 10.1007/s12539-015-0022-1
    In computed tomography (CT), blurring occurs due to different hardware or software errors and hides certain medical details that are present in an image. Image blur is difficult to avoid in many circumstances and can frequently ruin an image. For this, many methods have been developed to reduce the blurring artifact from CT images. The problems with these methods are the high implementation time, noise amplification and boundary artifacts. Hence, this article presents an amended version of the iterative Landweber algorithm to attain artifact-free boundaries and less noise amplification in a faster application time. In this study, both synthetic and real blurred CT images are used to validate the proposed method properly. Similarly, the quality of the processed synthetic images is measured using the feature similarity index, structural similarity and visual information fidelity in pixel domain metrics. Finally, the results obtained from intensive experiments and performance evaluations show the efficiency of the proposed algorithm, which has potential as a new approach in medical image processing.
  9. Adeshina AM, Hashim R
    Interdiscip Sci, 2016 Mar;8(1):53-64.
    PMID: 26260066 DOI: 10.1007/s12539-015-0274-9
    Stroke is a cardiovascular disease with high mortality and long-term disability in the world. Normal functioning of the brain is dependent on the adequate supply of oxygen and nutrients to the brain complex network through the blood vessels. Stroke, occasionally a hemorrhagic stroke, ischemia or other blood vessel dysfunctions can affect patients during a cerebrovascular incident. Structurally, the left and the right carotid arteries, and the right and the left vertebral arteries are responsible for supplying blood to the brain, scalp and the face. However, a number of impairment in the function of the frontal lobes may occur as a result of any decrease in the flow of the blood through one of the internal carotid arteries. Such impairment commonly results in numbness, weakness or paralysis. Recently, the concepts of brain's wiring representation, the connectome, was introduced. However, construction and visualization of such brain network requires tremendous computation. Consequently, previously proposed approaches have been identified with common problems of high memory consumption and slow execution. Furthermore, interactivity in the previously proposed frameworks for brain network is also an outstanding issue. This study proposes an accelerated approach for brain connectomic visualization based on graph theory paradigm using compute unified device architecture, extending the previously proposed SurLens Visualization and computer aided hepatocellular carcinoma frameworks. The accelerated brain structural connectivity framework was evaluated with stripped brain datasets from the Department of Surgery, University of North Carolina, Chapel Hill, USA. Significantly, our proposed framework is able to generate and extract points and edges of datasets, displays nodes and edges in the datasets in form of a network and clearly maps data volume to the corresponding brain surface. Moreover, with the framework, surfaces of the dataset were simultaneously displayed with the nodes and the edges. The framework is very efficient in providing greater interactivity as a way of representing the nodes and the edges intuitively, all achieved at a considerably interactive speed for instantaneous mapping of the datasets' features. Uniquely, the connectomic algorithm performed remarkably fast with normal hardware requirement specifications.
  10. Sajid MR, Muhammad N, Zakaria R, Shahbaz A, Bukhari SAC, Kadry S, et al.
    Interdiscip Sci, 2021 Jun;13(2):201-211.
    PMID: 33675528 DOI: 10.1007/s12539-021-00423-w
    BACKGROUND: In the broader healthcare domain, the prediction bears more value than an explanation considering the cost of delays in its services. There are various risk prediction models for cardiovascular diseases (CVDs) in the literature for early risk assessment. However, the substantial increase in CVDs-related mortality is challenging global health systems, especially in developing countries. This situation allows researchers to improve CVDs prediction models using new features and risk computing methods. This study aims to assess nonclinical features that can be easily available in any healthcare systems, in predicting CVDs using advanced and flexible machine learning (ML) algorithms.

    METHODS: A gender-matched case-control study was conducted in the largest public sector cardiac hospital of Pakistan, and the data of 460 subjects were collected. The dataset comprised of eight nonclinical features. Four supervised ML algorithms were used to train and test the models to predict the CVDs status by considering traditional logistic regression (LR) as the baseline model. The models were validated through the train-test split (70:30) and tenfold cross-validation approaches.

    RESULTS: Random forest (RF), a nonlinear ML algorithm, performed better than other ML algorithms and LR. The area under the curve (AUC) of RF was 0.851 and 0.853 in the train-test split and tenfold cross-validation approach, respectively. The nonclinical features yielded an admissible accuracy (minimum 71%) through the LR and ML models, exhibiting its predictive capability in risk estimation.

    CONCLUSION: The satisfactory performance of nonclinical features reveals that these features and flexible computational methodologies can reinforce the existing risk prediction models for better healthcare services.

  11. Mohammadi S, Parvizpour S, Razmara J, Abu Bakar FD, Illias RM, Mahadi NM, et al.
    Interdiscip Sci, 2018 Mar;10(1):157-168.
    PMID: 27475956 DOI: 10.1007/s12539-016-0180-9
    We report a detailed structural analysis of the psychrophilic exo-β-1,3-glucanase (GaExg55) from Glaciozyma antarctica PI12. This study elucidates the structural basis of exo-1,3-β-1,3-glucanase from this psychrophilic yeast. The structural prediction of GaExg55 remains a challenge because of its low sequence identity (37 %). A 3D model was constructed for GaExg55. Threading approach was employed to determine a suitable template and generate optimal target-template alignment for establishing the model using MODELLER9v15. The primary sequence analysis of GaExg55 with other mesophilic exo-1,3-β-glucanases indicated that an increased flexibility conferred to the enzyme by a set of amino acids substitutions in the surface and loop regions of GaExg55, thereby facilitating its structure to cold adaptation. A comparison of GaExg55 with other mesophilic exo-β-1,3-glucanases proposed that the catalytic activity and structural flexibility at cold environment were attained through a reduced amount of hydrogen bonds and salt bridges, as well as an increased exposure of the hydrophobic side chains to the solvent. A molecular dynamics simulation was also performed using GROMACS software to evaluate the stability of the GaExg55 structure at varying low temperatures. The simulation result confirmed the above findings for cold adaptation of the psychrophilic GaExg55. Furthermore, the structural analysis of GaExg55 with large catalytic cleft and wide active site pocket confirmed the high activity of GaExg55 to hydrolyze polysaccharide substrates.
  12. Elengoe A, Hamdan S
    Interdiscip Sci, 2017 Dec;9(4):478-498.
    PMID: 27517798 DOI: 10.1007/s12539-016-0181-8
    In this study, we explored the possibility of determining the synergistic interactions between nucleotide-binding domain (NBD) of Homo sapiens heat-shock 70 kDa protein (Hsp70) and E1A 32 kDa of adenovirus serotype 5 motif (PNLVP) in the efficiency of killing of tumor cells in cancer treatment. At present, the protein interaction between NBD and PNLVP motif is still unknown, but believed to enhance the rate of virus replication in tumor cells. Three mutant models (E229V, H225P and D230C) were built and simulated, and their interactions with PNLVP motif were studied. The PNLVP motif showed the binding energy and intermolecular energy values with the novel E229V mutant at -7.32 and -11.2 kcal/mol. The E229V mutant had the highest number of hydrogen bonds (7). Based on the root mean square deviation, root mean square fluctuation, hydrogen bonds, salt bridge, secondary structure, surface-accessible solvent area, potential energy and distance matrices analyses, it was proved that the E229V had the strongest and most stable interaction with the PNLVP motif among all the four protein-ligand complex structures. The knowledge of this protein-ligand complex model would help in designing Hsp70 structure-based drug for cancer therapy.
  13. Hon MK, Mohamad MS, Mohamed Salleh AH, Choon YW, Mohd Daud K, Remli MA, et al.
    Interdiscip Sci, 2019 Mar;11(1):33-44.
    PMID: 30758766 DOI: 10.1007/s12539-019-00324-z
    In recent years, metabolic engineering has gained central attention in numerous fields of science because of its capability to manipulate metabolic pathways in enhancing the expression of target phenotypes. Due to this, many computational approaches that perform genetic manipulation have been developed in the computational biology field. In metabolic engineering, conventional methods have been utilized to upgrade the generation of lactate and succinate in E. coli, although the yields produced are usually way below their theoretical maxima. To overcome the drawbacks  of such conventional methods, development of hybrid algorithm is introduced to obtain an optimal solution by proposing a gene knockout strategy in E. coli which is able to improve the production of lactate and succinate. The objective function of the hybrid algorithm is optimized using a swarm intelligence optimization algorithm and a Simple Constrained Artificial Bee Colony (SCABC) algorithm. The results maximize the production of lactate and succinate by resembling the gene knockout in E. coli. The Flux Balance Analysis (FBA) is integrated in a hybrid algorithm to evaluate the growth rate of E. coli as well as the productions of lactate and succinate. This results in the identification of a gene knockout list that contributes to maximizing the production of lactate and succinate in E. coli.
  14. Liu Y, Yu Y, Zhao S
    Interdiscip Sci, 2022 Jan 24.
    PMID: 35067893 DOI: 10.1007/s12539-021-00492-x
    LncRNAs play a part in numerous momentous processes of biology such as disease diagnoses, preventions and treatments. The associations between various diseases and lncRNAs are one of the crucial approaches to learn the role and status of lncRNAs in human diseases. With the researches on lncRNA and diseases, multiple methods based on neural network have been employed to predict these associations. However, the deep and complicated characteristic representations of lncRNA-disease associations were failed to be extracted, and the discriminative contributions of the interactions, correlations, and similarities among miRNAs diseases, and lncRNAs for the correlation predictions were ignored. In this paper, based on the multibiology premise of lncRNAs, miRNAs, and diseases, a dual attention network was proposed to predict the model of lncRNA-disease associations for miRNAs, the disease characteristic matrix, and lncRNAs. Through two attention modules, we enable the model to learn the nonlinear, more complex and useful features of lncRNA, miRNA, and disease characteristic matrix. For the feature embedding matrix composed of lncRNA-disease, the connection between lncRNA-disease feature embedding matrix and lncRNA, miRNA, and disease characteristic matrix was enhanced through deconvolution and feature fusion layer. Compared with several latest methods, the method proposed in this paper can produce better performance. Researches on the cases of osteosarcoma, lung cancer, and gastric cancer have confirmed the effective recognition of potential lncRNA-disease associations.
  15. Tan SL, Selvachandran G, Ding W, Paramesran R, Kotecha K
    Interdiscip Sci, 2023 Nov 14.
    PMID: 37962777 DOI: 10.1007/s12539-023-00589-5
    As one of the most common female cancers, cervical cancer often develops years after a prolonged and reversible pre-cancerous stage. Traditional classification algorithms used for detection of cervical cancer often require cell segmentation and feature extraction techniques, while convolutional neural network (CNN) models demand a large dataset to mitigate over-fitting and poor generalization problems. To this end, this study aims to develop deep learning models for automated cervical cancer detection that do not rely on segmentation methods or custom features. Due to limited data availability, transfer learning was employed with pre-trained CNN models to directly operate on Pap smear images for a seven-class classification task. Thorough evaluation and comparison of 13 pre-trained deep CNN models were performed using the publicly available Herlev dataset and the Keras package in Google Collaboratory. In terms of accuracy and performance, DenseNet-201 is the best-performing model. The pre-trained CNN models studied in this paper produced good experimental results and required little computing time.
  16. Goh YX, Wang M, Hou XP, He Y, Ou HY
    Interdiscip Sci, 2023 Sep;15(3):349-359.
    PMID: 36849628 DOI: 10.1007/s12539-023-00555-1
    The CRISPR‒Cas system acts as a bacterial defense mechanism by conferring adaptive immunity and limiting genetic reshuffling. However, under adverse environmental hazards, bacteria can employ their CRISPR‒Cas system to exchange genes that are vital for adaptation and survival. Levilactobacillus brevis is a lactic acid bacterium with great potential for commercial purposes because it can be genetically manipulated to enhance its functionality and nutritional value. Nevertheless, the CRISPR‒Cas system might interfere with the genetic modification process. Additionally, little is known about the CRISPR‒Cas system in this industrially important microorganism. Here, we investigate the prevalence, diversity, and targets of CRISPR‒Cas systems in the genus Levilactobacillus, further focusing on complete genomes of L. brevis. Using the CRISPRCasFinder webserver, we identified 801 putative CRISPR-Cas systems in the genus Levilactobacillus. Further investigation focusing on the complete genomes of L. brevis revealed 54 putative CRISPR-Cas systems. Of these, 46 were orphan CRISPRs, and eight were CRISPR‒Cas systems. The type II-A CRISPR‒Cas system is the most common in Levilactobacillus and L. brevis complete genomes. Analysis of the spacer's target showed that the CRISPR‒Cas systems of L. brevis mainly target the enterococcal plasmids. Comparative analysis of putative CRISPR-Cas loci in Levilactobacillus brevis.
  17. Peng P, Wu D, Huang LJ, Wang J, Zhang L, Wu Y, et al.
    Interdiscip Sci, 2023 Jul 24.
    PMID: 37486420 DOI: 10.1007/s12539-023-00580-0
    Breast cancer is commonly diagnosed with mammography. Using image segmentation algorithms to separate lesion areas in mammography can facilitate diagnosis by doctors and reduce their workload, which has important clinical significance. Because large, accurately labeled medical image datasets are difficult to obtain, traditional clustering algorithms are widely used in medical image segmentation as an unsupervised model. Traditional unsupervised clustering algorithms have limited learning knowledge. Moreover, some semi-supervised fuzzy clustering algorithms cannot fully mine the information of labeled samples, which results in insufficient supervision. When faced with complex mammography images, the above algorithms cannot accurately segment lesion areas. To address this, a semi-supervised fuzzy clustering based on knowledge weighting and cluster center learning (WSFCM_V) is presented. According to prior knowledge, three learning modes are proposed: a knowledge weighting method for cluster centers, Euclidean distance weights for unlabeled samples, and learning from the cluster centers of labeled sample sets. These strategies improve the clustering performance. On real breast molybdenum target images, the WSFCM_V algorithm is compared with currently popular semi-supervised and unsupervised clustering algorithms. WSFCM_V has the best evaluation index values. Experimental results demonstrate that compared with the existing clustering algorithms, WSFCM_V has a higher segmentation accuracy than other clustering algorithms, both for larger lesion regions like tumor areas and for smaller lesion areas like calcification point areas. Figure. The principle of Euclidean distance weighting for unlabeled samples.
  18. Ling L, Huang L, Wang J, Zhang L, Wu Y, Jiang Y, et al.
    Interdiscip Sci, 2023 Dec;15(4):560-577.
    PMID: 37160860 DOI: 10.1007/s12539-023-00570-2
    Soft subspace clustering (SSC), which analyzes high-dimensional data and applies various weights to each cluster class to assess the membership degree of each cluster to the space, has shown promising results in recent years. This method of clustering assigns distinct weights to each cluster class. By introducing spatial information, enhanced SSC algorithms improve the degree to which intraclass compactness and interclass separation are achieved. However, these algorithms are sensitive to noisy data and have a tendency to fall into local optima. In addition, the segmentation accuracy is poor because of the influence of noisy data. In this study, an SSC approach that is based on particle swarm optimization is suggested with the intention of reducing the interference caused by noisy data. The particle swarm optimization method is used to locate the best possible clustering center. Second, increasing the amount of geographical membership makes it possible to utilize the spatial information to quantify the link between different clusters in a more precise manner. In conclusion, the extended noise clustering method is implemented in order to maximize the weight. Additionally, the constraint condition of the weight is changed from the equality constraint to the boundary constraint in order to reduce the impact of noise. The methodology presented in this research works to reduce the amount of sensitivity the SSC algorithm has to noisy data. It is possible to demonstrate the efficacy of this algorithm by using photos with noise already present or by introducing noise to existing photographs. The revised SSC approach based on particle swarm optimization (PSO) is demonstrated to have superior segmentation accuracy through a number of trials; as a result, this work gives a novel method for the segmentation of noisy images.
Related Terms
Contact Us

Please provide feedback to Administrator (afdal@afpm.org.my)

External Links