Younghoon Kim has a total of 33 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Semi-Supervised Segmentation of Computed Tomography Images Using Co-Training of Models With Heterogeneous Tasks

    by Seunghan Yoon, Younghoon Kim

    Collecting sufficient training data for medical image segmentation models has long been a research challenge. To address this, various studies have employed semi-supervised learning (SSL), which combines labeled and unlabeled data to enhance model training. In abdominal CT images, segmentation errors often occur due to similar Hounsfield Unit values between target organs and surrounding tissues. This issue is particularly pronounced in environments with limited data. This poses a significant challenge in SSL research. Our study proposes an SSL method that leverages adjacent CT images and labeled data to incorporate 3D spatial information. An interpolation model generates pseudo-labels from the labels of adjacent CT data, which are used in co-training with a segmentation model to iteratively enhance performance. To evaluate performance, we conducted experiments on diverse abdominal CT datasets. The results demonstrate that the two models can improve each other’s outputs, achieving competitive performance even with limited training data.

  • Open Access

    Mitigating Quantization Errors Due to Activation Spikes in Gated Linear Unit-Based Large Language Models

    by Jaewoo Yang, Hayun Kim, Junyung Ji, Younghoon Kim

    Modern large language models (LLMs) achieve state-of-the-art performance through architectural advancements but require high computational costs for inference. Post-training quantization is a widely adopted approach to reduce these costs by quantizing weights and activations to lower precision, such as INT8. However, we identify a critical challenge in activation quantization for GLU (Gated Linear Unit) variants, which are commonly used in the feed-forward networks of modern LLMs like the LLaMA family. Specifically, severe local quantization errors arise due to excessively large activation magnitudes, which we refer to as activation spikes, leading to significant degradation in model performance. Our analysis reveals a systematic pattern of these spikes: they predominantly occur in the FFN (feed-forward network) layers at the early and late layers of the model and are concentrated on a small subset of tokens rather than being uniformly distributed across a token sequence. To mitigate this issue, we propose two empirical methods: Quantization-free Module (QFeM) and Quantization-free Prefix (QFeP), which isolate activation spikes during quantization. Extensive experiments demonstrated that our methods effectively improve activation quantization, particularly in coarse-grained quantization schemes, enhancing the performance of LLMs with GLU variants and addressing the limitations of existing quantization techniques. The code for implementing our methods and reproducing the experiments is publicly available our GitHub repository.


  • Open Access

    Accelerated muscle mass estimation from CT images through transfer learning

    by Seunghan Yoon, Tae Hyung Kim, Young Kul Jung, Younghoon Kim

    The cost of labeling to collect training data sets using deep learning is especially high in medical applications compared to other fields. Furthermore, due to variances in images depending on the computed tomography (CT) devices, a deep learning based segmentation model trained with a certain device often does not work with images from a different device. In this study, we propose an efficient learning strategy for deep learning models in medical image segmentation. We aim to overcome the difficulties of segmentation in CT images by training a VNet segmentation model which enables rapid labeling of organs in CT images with the model obtained by transfer learning using a small number of manually labeled images, called SEED images. We established a process for generating SEED images and conducting transfer learning a model. We evaluate the performance of various segmentation models such as vanilla UNet, UNETR, Swin-UNETR and VNet. Furthermore, assuming a scenario that a model is repeatedly trained with CT images collected from multiple devices, in which is catastrophic forgetting often occurs, we examine if the performance of our model degrades. We show that transfer learning can train a model that does a good job of segmenting muscles with a small number of images. In addition, it was confirmed that VNet shows better performance when comparing the performance of existing semi-automated segmentation tools and other deep learning networks to muscle and liver segmentation tasks. Additionally, we confirmed that VNet is the most robust model to deal with catastrophic forgetting problems. In the 2D CT image segmentation task, we confirmed that the CNN-based network shows better performance than the existing semi-automatic segmentation tool or latest transformer-based networks.


  • Open Access

    Predicting Missing Values in Survey Data Using Prompt Engineering for Addressing Item Non-Response

    by Junyung Ji, Jiwoo Kim, Younghoon Kim

    Survey data play a crucial role in various research fields, including economics, education, and healthcare, by providing insights into human behavior and opinions. However, item non-response, where respondents fail to answer specific questions, presents a significant challenge by creating incomplete datasets that undermine data integrity and can hinder or even prevent accurate analysis. Traditional methods for addressing missing data, such as statistical imputation techniques and deep learning models, often fall short when dealing with the rich linguistic content of survey data. These approaches are also hampered by high time complexity for training and the need for extensive preprocessing or feature selection. In this paper, we introduce an approach that leverages Large Language Models (LLMs) through prompt engineering for predicting item non-responses in survey data. Our method combines the strengths of both traditional imputation techniques and deep learning methods with the advanced linguistic understanding of LLMs. By integrating respondent similarities, question relevance, and linguistic semantics, our approach enhances the accuracy and comprehensiveness of survey data analysis. The proposed method bypasses the need for complex preprocessing and additional training, making it adaptable, scalable, and capable of generating explainable predictions in natural language. We evaluated the effectiveness of our LLM-based approach through a series of experiments, demonstrating its competitive performance against established methods such as Multivariate Imputation by Chained Equations (MICE), MissForest, and deep learning models like TabTransformer. The results show that our approach not only matches but, in some cases, exceeds the performance of these methods while significantly reducing the time required for data processing.


  • Open Access

    Adversarial Attacks on Medical Segmentation Model via Transformation of Feature Statistics

    by Woonghee Lee, Mingeon Ju, Yura Sim, Young Kul Jung, Tae Hyung Kim, Younghoon Kim

    Deep learning-based segmentation models have made a profound impact on medical procedures, with U-Net based computed tomography (CT) segmentation models exhibiting remarkable performance. Yet, even with these advances, these models are found to be vulnerable to adversarial attacks, a problem that equally affects automatic CT segmentation models. Conventional adversarial attacks typically rely on adding noise or perturbations, leading to a compromise between the success rate of the attack and its perceptibility. In this study, we challenge this paradigm and introduce a novel generation of adversarial attacks aimed at deceiving both the target segmentation model and medical practitioners. Our approach aims to deceive a target model by altering the texture statistics of an organ while retaining its shape. We employ a real-time style transfer method, known as the texture reformer, which uses adaptive instance normalization (AdaIN) to change the statistics of an image’s feature.To induce transformation, we modify the AdaIN, which typically aligns the source and target image statistics. Through rigorous experiments, we demonstrate the effectiveness of our approach. Our adversarial samples successfully pass as realistic in blind tests conducted with physicians, surpassing the effectiveness of contemporary techniques. This innovative methodology not only offers a robust tool for benchmarking and validating automated CT segmentation systems but also serves as a potent mechanism for data augmentation, thereby enhancing model generalization. This dual capability significantly bolsters advancements in the field of deep learning-based medical and healthcare segmentation models.


  • Open Access

    Enhancing CT Segmentation Security against Adversarial Attack: Most Activated Filter Approach

    by Woonghee Lee, Younghoon Kim

    This study introduces a deep-learning-based framework for detecting adversarial attacks in CT image segmentation within medical imaging. The proposed methodology includes analyzing features from various layers, particularly focusing on the first layer, and utilizing a convolutional layer-based model with specialized training. The framework is engineered to differentiate between tampered adversarial samples and authentic or noise-altered images, focusing on attack methods predominantly utilized in the medical sector. A significant aspect of the approach is employing a random forest algorithm as a binary classifier to detect attacks. This method has shown efficacy in identifying genuine samples and reducing false positives due to Gaussian noise. The contributions of this work include robust attack detection, layer-specific feature analysis, comprehensive evaluations, physician-friendly visualizations, and distinguishing between adversarial attacks and noise. This research enhances the security and reliability of CT image analysis in diagnostics.


  • Open Access

    Pixel Diffuser: Practical Interactive Medical Image Segmentation without Ground Truth

    by Mingeon Ju, Jaewoo Yang, Jaeyoung Lee, Moonhyun Lee, Junyung Ji, Younghoon Kim

    Medical image segmentation is essential for doctors to diagnose diseases and manage patient status. While deep learning has demonstrated potential in addressing segmentation challenges within the medical domain, obtaining a substantial amount of data with accurate ground truth for training high-performance segmentation models is both time-consuming and demands careful attention. While interactive segmentation methods can reduce the costs of acquiring segmentation labels for training supervised models, they often still necessitate considerable amounts of ground truth data. Moreover, achieving precise segmentation during the refinement phase results in increased interactions. In this work, we propose an interactive medical segmentation method called PixelDiffuser that requires no medical segmentation ground truth data and only a few clicks to obtain high-quality segmentation using a VGG19-based autoencoder. As the name suggests, PixelDiffuser starts with a small area upon the initial click and gradually detects the target segmentation region. Specifically, we segment the image by creating a distortion in the image and repeating it during the process of encoding and decoding the image through an autoencoder. Consequently, PixelDiffuser enables the user to click a part of the organ they wish to segment, allowing the segmented region to expand to nearby areas with pixel values similar to the chosen organ. To evaluate the performance of PixelDiffuser, we employed the dice score, based on the number of clicks, to compare the ground truth image with the inferred segment. For validation of our method’s performance, we leveraged the BTCV dataset, containing CT images of various organs, and the CHAOS dataset, which encompasses both CT and MRI images of the liver, kidneys and spleen. Our proposed model is an efficient and effective tool for medical image segmentation, achieving competitive performance compared to previous work in less than five clicks and with very low memory consumption without additional training.


  • Open Access

    MT-VTON: Multilevel Transformation-Based Virtual Try-On for Enhancing Realism of Clothing

    by Jaeyoung Lee, Moonhyun Lee, Younghoon Kim

    Virtual 3D fashion fitting, commonly referred to as 2D virtual try-on, has garnered significant attention due to its potential to revolutionize the way consumers interact with fashion items online. This paper presents a novel approach to virtual try-on utilizing a deep learning framework built upon the concept of appearance flow. Our proposed method improves the existing state-of-the-art techniques by seamlessly integrating natural cloth folds, shadows, and intricate textures, such as letters and comic characters, into the synthesized virtual try-on images. Building upon the advancements of previous research, our approach introduces a multi-faceted transformation strategy that operates at both the pixel and image patch levels. Our method’s effectiveness is demonstrated through extensive experiments and comparisons with existing virtual try-on techniques. The results showcase a substantial improvement in the synthesis of virtual try-on images with natural-looking cloth folds, realistic shadows, and intricate textures.


  • Open Access

    Automated Clinical Impression Generation for Medical Signal Data Searches

    by Woonghee Lee, Jaewoo Yang, Doyeong Park, Younghoon Kim

    Medical retrieval systems have become significantly important in clinical settings. However, commercial retrieval systems that heavily rely on term-based indexing face challenges when handling continuous medical data, such as electroencephalography data, primarily due to the high cost associated with utilizing neurologist analyses. With the increasing affordability of data recording systems, it becomes increasingly crucial to address these challenges. Traditional procedures for annotating, classifying, and interpreting medical data are costly, time consuming, and demand specialized knowledge. While cross-modal retrieval systems have been proposed to address these challenges, most concentrate on images and text, sidelining time-series medical data like electroencephalography data. As the interpretation of electroencephalography signals, which document brain activity, requires a neurologist’s expertise, this process is often the most expensive component. Therefore, a retrieval system capable of using text to identify relevant signals, eliminating the need for expert analysis, is desirable. Our research proposes a solution to facilitate the creation of indexing systems employing electroencephalography signals for report generation in situations where reports are pending a neurologist review. We introduce a method incorporating a convolutional-neural-network-based encoder from DeepSleepNet, which extracts features from electroencephalography signals, coupled with a transformer which learns the signal’s auto-correlation and the relationship between the signal and the corresponding report. Experimental evaluation using real-world data revealed our approach surpasses baseline methods. These findings suggest potential advancements in medical data retrieval and a decrease in reliance on expert knowledge for electroencephalography signal analysis. As such, our research represents a significant stride towards making electroencephalography data more comprehensible and utilizable in clinical environments.


  • Open Access

    Model Interpretation Considering Both Time and Frequency Axes Given Time Series Data

    by Woonghee Lee, Gayeon Kim, Jeonghyeon Yu, Younghoon Kim

    Recently, deep learning-based models have emerged in the medical domain. Although those models achieve high performance, it is difficult to directly apply them in practice. Specifically, most models are not considered reliable yet, while they are not interpretable. Therefore, researchers attempt to interpret their own deep learning applications. However, the interpretation is task specific or only appropriate for image data such as computed tomography (CT) scans and magnetic resonance imaging (MRI). Currently, few works focus on the model interpretation given time series data such as electroencephalography (EEG) and electrocardiography (ECG) using LIME. Because the explanation generated by LIME is from the permutation of the divided input data, the performance of interpretation is highly dependent on the split method. In the medical domain, for the time series data, existing interpretations consider only the time axis, whereas physicians take account of the frequency too. In this work, we propose the model interpretation using LIME considering both time and frequency axes. Our key idea is that we divide the input signal using graph-based image clustering after transforming it using short-time Fourier transform, which is utilized to capture the change of frequency content over time. In our experiments, we utilize real-world data, which is EEG signals recorded from patients during polysomnographic (PSG) studies, as well as prove that ours captures a significantly more critical explanation than the state-of-the-art. In addition, we show that the representation obtained by ours reflects the physician’s standard such as K-complexes and delta waves, which are considered strong evidence of the second sleep stage and a clue of the third sleep stage. We expect that our work can be applied to establish computer-aided diagnosis as well as to measure the reliability of deep learning models taking the time series into them.