In this paper, we focus on the problem of Medical Visual Question Answering (MedVQA), which is crucial in efficiently interpreting medical images with vital clinic-relevant information. Firstly, reframe MedVQA as a generation task that naturally follows human-machine interaction, propose generative-based model for visual understanding by aligning information from pre-trained vision encoder large language model. Secondly, establish scalable pipeline to construct large-scale question-answering...

Driven by the large foundation models, development of artificial intelligence has witnessed tremendous progress lately, leading to a surge general interest from public. In this study, we aim assess performance OpenAI's newest model, GPT-4V(ision), specifically in realm multimodal medical diagnosis. Our evaluation encompasses 17 human body systems, including Central Nervous System, Head and Neck, Cardiac, Chest, Hematology, Hepatobiliary, Gastrointestinal, Urogenital, Gynecology, Obstetrics,...

The timely detection and segmentation of pulmonary nodules in lung computed tomography (CT) images can aid the early diagnosis treatment cancer. However, manual by doctors is highly demanding terms operational requirements efficiency. To effectively improve nodule segmentation, this paper proposes a novel neural network, called ResDSda_U-Net, based on original U-Net network with following improvements: (1) combining Depthwise Over-parameterized Convolutional layer (DO-Conv) simple...

In order to enable the calibration model be effectively transferred among multiple instruments and correct differences between spectra measured by different instruments, a new feature transfer based on partial least squares regression (PLS) subspace (PLSCT) is proposed in this paper. Firstly, PLS of master instrument built, meanwhile constructed vectors. Then slave are projected into subspace, features also extracted at same time. pseudo predicted ordinary method so that it matches spectra....

Calibration transfer is an important field for near-infrared (NIR) spectroscopy in practical applications. However, most methods are constructed with standard samples, which expensive and difficult to obtain. Taking this problem into account, paper proposes a calibration method based on affine invariance without standards (CTAI). Our can be utilized adjust the difference between two instruments by transformation. CTAI firstly establishes partial least squares (PLS) model of master instrument...

In this study, we focus on building up a model that can Segment Anything in medical scenarios, driven by Text prompts, termed as SAT. Our main contributions are three folds: (i) data construction, combine multiple knowledge sources to construct multi-modal tree; Then build large-scale segmentation dataset for training, collecting over 11K 3D image scans from 31 datasets with careful standardization both visual and label space; (ii) formulate universal model, be prompted inputting...

Developing generalist foundation model has recently attracted tremendous attention among researchers in the field of AI for Medicine (AI4Medicine). A pivotal insight developing these models is their reliance on dataset scaling, which emphasizes requirements open-source medical image datasets that incorporate diverse supervision signals across various imaging modalities. In this paper, we introduce RadGenome-Chest CT, a comprehensive, large-scale, region-guided 3D chest CT interpretation...

Medical Visual Question Answering (MedVQA) enhances diagnostic accuracy and healthcare delivery by leveraging artificial intelligence to interpret medical images. This study aims redefine MedVQA as a generation task that mirrors human–machine interaction develop model capable of integrating complex visual textual information. We constructed large-scale visual-question answering dataset, PMC-VQA, containing 227,000 VQA pairs across 149,000 images span various modalities diseases. introduced...

Foundation models trained on large-scale dataset gain a recent surge in CV and NLP. In contrast, development biomedical domain lags far behind due to data scarcity. To address this issue, we build release PMC-OA, with 1.6M image-caption pairs collected from PubMedCentral's OpenAccess subset, which is 8 times larger than before. PMC-OA covers diverse modalities or diseases, majority of the samples aligned at finer-grained level, i.e., subfigure subcaption. While pretraining CLIP-style model...

Medical image segmentation has recently demonstrated impressive progress with deep neural networks, yet the heterogeneous modalities and scarcity of mask annotations limit development models on unannotated modalities. This paper investigates a new paradigm for leveraging generative in medical applications: controllably synthesizing data modalities, without requiring registered pairs. Specifically, we make following contributions this paper: (i) collect curate large-scale radiology image-text...

Postoperative recurrence was a life-threatening condition for patients with rectal cancer. Due to the heterogeneity of locally recurrent cancer (LRRC) and controversy optimal treatment patients, it difficult predict prognosis LRRC. This study aimed develop validate nomogram that could accurately survival probability LRRC.Patients diagnosed LRRC between 2004 2019 from Surveillance, Epidemiology, End Results (SEER) database were included in analysis. Multiple imputations chained equations used...

In recent years, along with the further study in incomplete information chess-military chess. How to express and address during process of game, deciding certain moves strategies obtain higher winning rate increasingly become a new issue. An approach using digit military chess is proposed this paper, model based on guessing probability game designed, programming provided. National Computer Games Tournament, program stronger than others. Results are shown as 6 wins, 1 draw, lost. The average...

Computer game is a classic research field of Artificial intelligence, its history almost as long intelligence. Military chess kind computer incomplete information and position pieces are uncharted. Game trees' branching factor military huge searching space big. So the method tree isn't used in this paper. The probability table algorithm adopted Firstly, basic steps generated algorithm, then current state can be assessed found timely, at last, superiority proved national competition.

This paper considers the problem of undersampled MRI reconstruction. We propose a novel Transformer-based framework for directly processing signal in k-space, going beyond limitation regular grids as ConvNets do. adopt an implicit representation k-space spectrogram, treating spatial coordinates inputs, and dynamically query sparsely sampled points to reconstruct i.e. learning inductive bias k-space. To strike balance between computational cost reconstruction quality, we build decoder with...

This study first presents a taxonomy of region-based and regression-based target identification methods, then it evaluates each individual technique in turn. The former can extract picture characteristics from the CNN backbone apply sliding window procedures to candidate areas lessen computational complexity. latter immediately takes data convolution layer for feature extraction, classification, location regression rather than going via intermediate retrieve areas. Although accuracy is only...