Weight Pruning-UNet

Rao, Patike Kiran; Chatterjee, Subarna; Sharma, Sreedhar

doi:10.4103/jmss.jmss_108_21

Cited by 3 publications

(6 citation statements)

References 14 publications

(15 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The methodology of the proposed work unfolds in two distinct phases, underpinned by a decoupled deep learning model approach designed for the reconstruction of 3D medical images from 2D slices and the subsequent segmentation of anatomical structures such as kidneys and kidney tumors. In Phase I, titled "SculptorGAN," a novel Conditional Generative Adversarial Network (cGAN) integrated with a Weight Pruning U-Net (WP-UNet) architecture [33] is employed to enhance, interpolate, and assemble 2D slices from the KiTs19 dataset into coherent 3D volumes. This phase leverages the conditional generative capabilities of SculptorGAN to ensure the contextual and spatial continuity of the reconstructed 3D volumes, optimizing the process through weight pruning techniques for computational efficiency.…”

Section: Methodsmentioning

confidence: 99%

“…This phase leverages the conditional generative capabilities of SculptorGAN to ensure the contextual and spatial continuity of the reconstructed 3D volumes, optimizing the process through weight pruning techniques for computational efficiency. Phase II advances with the segmentation task, utilizing a separate WP-UNet model [33] specifically tuned for the segmentation of kidneys and kidney tumors within the 3D volumes generated by SculptorGAN. This phase emphasizes precision in voxelwise classification, employing weight-pruned networks to maintain model performance while reducing computational load.…”

Section: Methodsmentioning

confidence: 99%

“…The SculptorGAN algorithm, as outlined in Algorithm 1, delineates a comprehensive approach for reconstructing 3D medical images from a sequence of 2D slices, specifically tailored to the needs of medical imaging analysis within the KiTs19 dataset. This method capitalizes on the capabilities of a Conditional Generative Adversarial Network (cGAN) [34] integrated with a Weight Pruning U-Net (WP-UNet) [33] architecture, designed to enhance, interpolate, and assemble 2D medical slices into volumetric 3D representations. The process commences with preprocessing steps, including normalization of intensity values and data augmentation, to prepare the 2D slices for processing.…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Decoupled SculptorGAN Framework for 3D Reconstruction and Enhanced Segmentation of Kidney Tumors in CT Images

Prakash,

Rao,

Babu

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Our proposed work, SculptorGAN, represents a novel advancement in the domain of medical imaging, for the accurate and automatic diagnosis of renal tumors, using the techniques and principles of Generative Adversarial Network (GAN). This dichotomous framework forms a contrast to the normal segmentation models like that of U-Net model but, instead, founded on a strategy that is aimed towards reconstruction and segmentation of CT images, particularly of renal malignancies. The core of the SculptorGAN methodology is a GAN-based approach for precise three-dimensional rendering of renal anatomies from CT scans, followed by a segmentation phase to correctly separate the neoplastic from non-neoplastic tissues. In fact, SculptorGAN was designed to circumvent limitations that come as inherent in the segmentation techniques, and in this case to eliminate them. In fact, by including such an advanced algorithmic architecture, accuracy of diagnosis in SculptorGAN has increased to 96.5%, which is the primary aspect behind early detection and thus proper curing of renal tumors. The better results were ascribed to more accurate and detailed reconstruction of renal structures that the framework allowed, apart from the better segmentation. The performance analyses show quantitative results with respect to the presented datasets, while the validation shows that SculptorGAN outperforms most of the traditional models such as U-Net. In particular, SculptorGAN decreased the time taken for 3D reconstruction by about 35% while increasing the accuracy of segmentation by 20% or more. The outcome, in their turn, may suggest this improvement in efficiency and the level of reliability for renal tumor diagnosis as of having farreaching implications for the patient treatment and its outcomes. In conclusion, the framework deals with all the challenges with an accurate diagnosis of renal tumors and brings betterment in the overall field of medical image analysis by providing the abilities of GANs for the betterment in image reconstruction and segmentation.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Decoupled SculptorGAN Framework for 3D Reconstruction and Enhanced Segmentation of Kidney Tumors in CT Images

Prakash,

Rao,

Babu

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…We divided these datasets into training, validation, and test sets, with 342 cases allocated for training, 73 cases for validation, and 73 for testing. We preprocessed the datasets to ensure consistent input dimensions and normalized pixel values [43].…”

Section: Methodsmentioning

confidence: 99%

“…To enhance submodel efficiency, we employ a technique known as weight pruning, as referenced in [43][44][45]. This involves selectively reducing the number of parameters within a submodel by setting specific weight values to zero.…”

Section: Weight Pruning For Efficient Resource Utilizationmentioning

confidence: 99%

Optimizing Inference Distribution for Efficient Kidney Tumor Segmentation Using a UNet-PWP Deep-Learning Model with XAI on CT Scan Images

Rao,

Chatterjee,

Janardhan

et al. 2023

Diagnostics

Self Cite

View full text Add to dashboard Cite

Kidney tumors represent a significant medical challenge, characterized by their often-asymptomatic nature and the need for early detection to facilitate timely and effective intervention. Although neural networks have shown great promise in disease prediction, their computational demands have limited their practicality in clinical settings. This study introduces a novel methodology, the UNet-PWP architecture, tailored explicitly for kidney tumor segmentation, designed to optimize resource utilization and overcome computational complexity constraints. A key novelty in our approach is the application of adaptive partitioning, which deconstructs the intricate UNet architecture into smaller submodels. This partitioning strategy reduces computational requirements and enhances the model’s efficiency in processing kidney tumor images. Additionally, we augment the UNet’s depth by incorporating pre-trained weights, therefore significantly boosting its capacity to handle intricate and detailed segmentation tasks. Furthermore, we employ weight-pruning techniques to eliminate redundant zero-weighted parameters, further streamlining the UNet-PWP model without compromising its performance. To rigorously assess the effectiveness of our proposed UNet-PWP model, we conducted a comparative evaluation alongside the DeepLab V3+ model, both trained on the “KiTs 19, 21, and 23” kidney tumor dataset. Our results are optimistic, with the UNet-PWP model achieving an exceptional accuracy rate of 97.01% on both the training and test datasets, surpassing the DeepLab V3+ model in performance. Furthermore, to ensure our model’s results are easily understandable and explainable. We included a fusion of the attention and Grad-CAM XAI methods. This approach provides valuable insights into the decision-making process of our model and the regions of interest that affect its predictions. In the medical field, this interpretability aspect is crucial for healthcare professionals to trust and comprehend the model’s reasoning.

show abstract

MTAN: A semi-supervised learning model for kidney tumor segmentation

Sun,

Yang,

Guan

et al. 2023

XST

View full text Add to dashboard Cite

BACKGROUND: Medical image segmentation is crucial in disease diagnosis and treatment planning. Deep learning (DL) techniques have shown promise. However, optimizing DL models requires setting numerous parameters, and demands substantial labeled datasets, which are labor-intensive to create. OBJECTIVE: This study proposes a semi-supervised model that can utilize labeled and unlabeled data to accurately segment kidneys, tumors, and cysts on CT images, even with limited labeled samples. METHODS: An end-to-end semi-supervised learning model named MTAN (Mean Teacher Attention N-Net) is designed to segment kidneys, tumors, and cysts on CT images. The MTAN model is built on the foundation of the AN-Net architecture, functioning dually as teachers and students. In its student role, AN-Net learns conventionally. In its teacher role, it generates objects and instructs the student model on their utilization to enhance learning quality. The semi-supervised nature of MTAN allows it to effectively utilize unlabeled data for training, thus improving performance and reducing overfitting. RESULTS: We evaluate the proposed model using two CT image datasets (KiTS19 and KiTS21). In the KiTS19 dataset, MTAN achieved segmentation results with an average Dice score of 0.975 for kidneys and 0.869 for tumors, respectively. Moreover, on the KiTS21 dataset, MTAN demonstrates its robustness, yielding average Dice scores of 0.977 for kidneys, 0.886 for masses, 0.861 for tumors, and 0.759 for cysts, respectively. CONCLUSION: The proposed MTAN model presents a compelling solution for accurate medical image segmentation, particularly in scenarios where the labeled data is scarce. By effectively utilizing the unlabeled data through a semi-supervised learning approach, MTAN mitigates overfitting concerns and achieves high-quality segmentation results. The consistent performance across two distinct datasets, KiTS19 and KiTS21, underscores model’s reliability and potential for clinical reference.

show abstract

Weight Pruning-UNet

Cited by 3 publications

References 14 publications

Decoupled SculptorGAN Framework for 3D Reconstruction and Enhanced Segmentation of Kidney Tumors in CT Images

Decoupled SculptorGAN Framework for 3D Reconstruction and Enhanced Segmentation of Kidney Tumors in CT Images

Optimizing Inference Distribution for Efficient Kidney Tumor Segmentation Using a UNet-PWP Deep-Learning Model with XAI on CT Scan Images

MTAN: A semi-supervised learning model for kidney tumor segmentation

Contact Info

Product

Resources

About