Search

While computer vision has achieved tremendous success with multimodal encoding and direct textual interaction with images via chat-based large language models, similar advancements in medical imaging AI, particularly in 3D imaging, have been limited due to the scarcity of comprehensive datasets. To address this critical gap, we introduce CT-RATE, the first dataset that pairs 3D medical images with corresponding textual reports. CT-RATE comprises 25,692 non-contrast 3D chest CT scans from 21,304 unique patients. Through various reconstructions, these scans are expanded to 50,188 volumes, totaling over 14.3 million 2D slices. Each scan is accompanied by its corresponding radiology report. Leveraging CT-RATE, we develop CT-CLIP, a CT-focused contrastive language-image pretraining framework designed for broad applications without the need for task-specific training. We demonstrate how CT-CLIP can be used in two tasks: multi-abnormality detection and case retrieval. Remarkably, in multi-abnormality detection, CT-CLIP outperforms state-of-the-art fully supervised models across all key metrics, effectively eliminating the need for manual annotation. In case retrieval, it efficiently retrieves relevant cases using either image or textual queries, thereby enhancing knowledge dissemination. By combining CT-CLIP's vision encoder with a pretrained large language model, we create CT-CHAT, a vision-language foundational chat model for 3D chest CT volumes. Finetuned on over 2.7 million question-answer pairs derived from the CT-RATE dataset, CT-CHAT surpasses other multimodal AI assistants, underscoring the necessity for specialized methods in 3D medical imaging. Collectively, the open-source release of CT-RATE, CT-CLIP, and CT-CHAT not only addresses critical challenges in 3D medical imaging but also lays the groundwork for future innovations in medical AI and improved patient care.

GenerateCT, the first approach to generating 3D medical imaging conditioned on free-form medical text prompts, incorporates a text encoder and three key components: a novel causal vision transformer for encoding 3D CT volumes, a text-image transformer for aligning CT and text tokens, and a text-conditional super-resolution diffusion model. Without directly comparable methods in 3D medical imaging, we benchmarked GenerateCT against cutting-edge methods, demonstrating its superiority across all key metrics. Importantly, we evaluated GenerateCT's clinical applications in a multi-abnormality classification task. First, we established a baseline by training a multi-abnormality classifier on our real dataset. To further assess the model's generalization to external data and performance with unseen prompts in a zero-shot scenario, we employed an external set to train the classifier, setting an additional benchmark. We conducted two experiments in which we doubled the training datasets by synthesizing an equal number of volumes for each set using GenerateCT. The first experiment demonstrated an 11% improvement in the AP score when training the classifier jointly on real and generated volumes. The second experiment showed a 7% improvement when training on both real and generated volumes based on unseen prompts. Moreover, GenerateCT enables the scaling of synthetic training datasets to arbitrary sizes. As an example, we generated 100,000 3D CTs, fivefold the number in our real set, and trained the classifier exclusively on these synthetic CTs. Impressively, this classifier surpassed the performance of the one trained on all available real data by a margin of 8%. Last, domain experts evaluated the generated volumes, confirming a high degree of alignment with the text prompt. Access our code, model weights, training data, and generated data at https://github.com/ibrahimethemhamamci/GenerateCT

In this study, the four joint time-frequency (TF) moments; mean, variance, skewness, and kurtosis of TF matrix obtained from Multivariate Synchrosqueezing Transform (MSST) are proposed as features for hand gesture recognition. A publicly available dataset containing surface EMG (sEMG) signals of 40 subjects performing 10 hand gestures, was used. The distinguishing power of the feature variables for the tested gestures was evaluated according to their p values obtained from the Kruskal-Wallis (KW) test. It is concluded that the mean, variance, skewness, and kurtosis of TF matrices can be candidate feature sets for the recognition of hand gestures., Comment: Conference Paper, 5 pages, Translated Title (TR): El Hareketi Siniflandirmasi icin Cok-Degiskenli Senkrosikistirma Donusumune Dayali Zaman-Frekans Ozniteliklerinin Istatistiksel Analizi Proceedings: https://www.tipcih.com/books/ICMD2022_Proceedings-DRAFT.pdf Website: https://www.tipcih.com/

EEG signals are complex and low-frequency signals. Therefore, they are easily influenced by external factors. EEG artifact removal is crucial in neuroscience because artifacts have a significant impact on the results of EEG analysis. The removal of ocular artifacts is the most challenging among these artifacts. In this study, a novel ocular artifact removal method is presented by developing bidirectional long-short term memory (BiLSTM)-based deep learning (DL) models. We created a benchmarking dataset to train and test proposed DL models by combining the EEGdenoiseNet and DEAP datasets. We also augmented the data by contaminating ground-truth clean EEG signals with EOG at various SNR levels. The BiLSTM network is then fed to features extracted from augmented signals using highly-localized time-frequency (TF) coefficients obtained by wavelet synchrosqueezed transform (WSST). We also compare the WSST-based DL model results with traditional TF analysis (TFA) methods namely short-time Fourier transformation (STFT) and continuous wavelet transform (CWT) as well as augmented raw signals. The best average MSE value of 0.3066 was obtained by the first time-proposed BiLSTM-based WSST-Net model. Our results demonstrated the WSST-Net model significantly improves artifact removal performance compared to traditional TF and raw signal methods. Also, the proposed EOG removal approach reveals that it outperforms many conventional and DL-based ocular artifact removal methods in the literature., Comment: 6 pages, 3 figures

Plasma is defined as the fourth state of matter and non-thermal plasma can be produced at atmospheric pressure under a high electrical field. The strong and broad-spectrum antimicrobial effect of plasma-activated liquids (PALs) is now well known. The proven applicability of machine learning (ML) in the medical field is encouraging for its application in the field of plasma medicine as well. Thus, ML applications on PALs could present a new perspective to better understand the influences of various parameters on their antimicrobial effects. In this paper, comparative supervised ML models are presented by using previously obtained data to qualitatively predict the in vitro antimicrobial activity of PALs. A literature search was performed and data is collected from 33 relevant articles. After the required preprocessing steps, two supervised ML methods, namely classification, and regression are applied to data to obtain microbial inactivation (MI) predictions. For classification, MI is labeled in four categories and for regression, MI is used as a continuous variable. Two different robust cross-validation strategies are conducted for classification and regression models to evaluate the proposed method; repeated stratified k-fold cross-validation and k-fold cross-validation, respectively. We also investigate the effect of different features on models. The results demonstrated that the hyperparameter-optimized Random Forest Classifier (oRFC) and Random Forest Regressor (oRFR) provided better results than other models for the classification and regression, respectively. Finally, the best test accuracy of 82.68% for oRFC and R2 of 0.75 for the oRFR are obtained. ML techniques could contribute to a better understanding of plasma parameters that have a dominant role in the desired antimicrobial effect. Furthermore, such findings may contribute to the definition of a plasma dose in the future., Comment: 45 pages, 11 figures, 5 tables

The postoperative range of motion is one of the crucial factors indicating the outcome of Total Knee Arthroplasty (TKA). Although the correlation between range of knee flexion and posterior condylar offset (PCO) is controversial in the literature, PCO maintains its importance on evaluation of TKA. Due to limitations on PCO measurement, two novel parameters, posterior condylar offset ratio (PCOR) and anterior condylar offset ratio (ACOR), were introduced. Nowadays, the calculation of PCOR and ACOR on plain lateral radiographs is done manually by orthopedic surgeons. In this regard, we developed a software, AutoCOR, to calculate PCOR and ACOR autonomously, utilizing unsupervised machine learning algorithm (k-means clustering) and digital image processing techniques. The software AutoCOR is capable of detecting the anterior/posterior edge points and anterior/posterior cortex of the femoral shaft on true postoperative lateral conventional radiographs. To test the algorithm, 50 postoperative true lateral radiographs from Istanbul Kosuyolu Medipol Hospital Database were used (32 patients). The mean PCOR was 0.984 (SD 0.235) in software results and 0.972 (SD 0.164) in ground truth values. It shows strong and significant correlation between software and ground truth values (Pearson r=0.845 p<0.0001). The mean ACOR was 0.107 (SD 0.092) in software results and 0.107 (SD 0.070) in ground truth values. It shows moderate and significant correlation between software and ground truth values (Spearman's rs=0.519 p=0.0001412). We suggest that AutoCOR is a useful tool that can be used in clinical practice., Comment: 9 pages

As FM coverage is so ubiquitous around the world, several applications can be considered to better exploit this useful band. Thus, it is of significant interest to investigate and characterize channel properties of the FM Band for the potential two-way digital wireless systems. In this paper, we present the results of field measurements at 86 MHz conducted at Gebze, Kocaeli, Turkey. Through the measurements, some of the FM channel characteristics are identified. Measurements are performed for urban, hilly terrain, and rural areas. Our results show that the FM channel expectedly has a large coverage area but at the same time, it possesses large channel excess delays. While most of COST-207 channel power delay profile models are also applicable for the FM Band, for urban environments and hilly terrain environments, channel clusters and excess delays are higher than those of COST-207 models. As 5G systems aim to utilize lower frequency bands for supplementary links, FM Band can be considered as one of the potential bands and the channel models proposed in this study can then be exploited for performance analysis.

Multi-armed bandits (MAB) are extensively studied in various settings where the objective is to \textit{maximize} the actions' outcomes (i.e., rewards) over time. Since safety is crucial in many real-world problems, safe versions of MAB algorithms have also garnered considerable interest. In this work, we tackle a different critical task through the lens of \textit{linear stochastic bandits}, where the aim is to keep the actions' outcomes close to a target level while respecting a \textit{two-sided} safety constraint, which we call \textit{leveling}. Such a task is prevalent in numerous domains. Many healthcare problems, for instance, require keeping a physiological variable in a range and preferably close to a target level. The radical change in our objective necessitates a new acquisition strategy, which is at the heart of a MAB algorithm. We propose SALE-LTS: Safe Leveling via Linear Thompson Sampling algorithm, with a novel acquisition strategy to accommodate our task and show that it achieves sublinear regret with the same time and dimension dependence as previous works on the classical reward maximization problem absent any safety constraint. We demonstrate and discuss our algorithm's empirical performance in detail via thorough experiments., Comment: 17 pages, 4 figures

In wireless communication systems, identifying the number of channel taps offers an enhanced estimation of the channel impulse response (CIR). In this work, efficient identification of the number of wireless channel taps has been achieved via deep neural networks (DNNs), where we modified an existing DNN and analyzed its convergence performance using only the transmitted and received signals of a wireless system. The displayed results demonstrate that the adopted DNN accomplishes superior performance in identifying the number of channel taps, as compared to an existing algorithm called Spectrum Weighted Identification of Signal Sources (SWISS).

Positron emission tomography (PET) is a medical imaging method based on the measurement of concentrations of positron-emitting radionuclides in a living body. In the PET imaging system, glucose is labeled with a positron-emitting radionuclide and injected intravenously. Then, the positrons move through the tissue and collide with the electrons of the cells in which they interact. As a result of this interaction, two gamma rays are emitted in the opposite direction. Gama rays emitted from cancerous tissue that has retained radioactive glucose are detected through ring-shaped detectors. And the detected signals are converted into an electrical response. Subsequently, these responses are sampled with electronic circuits and recorded as histogram matrix to generate the image set. The gamma rays may not reach the detectors located in the opposite position in equal time. In PETs having TOF characteristics, it is aimed to obtain better positioning information by a method based on the principle of measuring the difference between the reach time of the two photons to detectors. The measurement of the flight time is carried out with TDC structures. The measurement of this time difference at the ps level is directly related to the spatial resolution of the PET system. In this study, 45 nm CMOS VLSI simulations of TDC structures that have various architectural approaches were performed for use in PET systems. With the designed TDC architectures, two gamma photons time reach to detectors have been simulated and the time difference has been successfully digitized. In addition, various performance metrics such as input and output voltages, time resolutions, measurement ranges, and power analysis of TDC architectures have been determined. Proposed Vernier oscillator-based TDC architecture has been reached 25 ps time resolution with a low power consumption of 1.62681 mW at 1V supply voltage., Comment: 8 pages, in Turkish language. 6 figures, conference paper,International Marmara Sciences Congess (IMASCON 2019 SPRING), https://www.imascon.com/dosyalar/imascon2019bahar/imascon_fen_bildiriler_ciltII_bahar_2019.pdf , https://avesis.kocaeli.edu.tr/yayin/99073ee1-45ff-495e-9cab-42de4d0fad71/vlsi-implementation-of-tdc-architectures-used-in-pet-imaging-systems

In this study, we describe the behavior of LTE over the sea and investigate the problem of radio resource block allocation in such SINR limited maritime channels. For simulations of such sea environment, we considered a network scenario of Bosphorus Strait in Istanbul, Turkey with different number of ships ferrying between two ports at a given time. After exploiting the network characteristics, we formulated and solved the radio resource allocation problem by max-min integer linear programming method. The radio resource allocation fairness in terms of Jain's fairness index was computed and it was compared with round robin and opportunistic methods. Results show that the max-min optimization method performs better than the opportunistic and round robin methods. This result in turn reflects that the max-min optimization method gives us the high minimum best throughput as compared to other two methods considering different ship density scenarios in the sea. Also, it was observed that as the number of ships begin to increase in the sea, the max-min method performs significantly better with good fairness as compared to the other two methods., Comment: 6 pages, 10 figures. Published in 2016 IEEE 37th Sarnoff Symposium at Newark, NJ, USA