TipDet: A multi-keyframe motion-aware framework for tip detection during ultrasound-guided interventions.

BACKGROUND AND OBJECTIVE: Automatic needle tip detection is important in real-time ultrasound (US) images that are utilized to guide interventional needle puncture procedures in clinical settings. However, due to the spatial indiscernibility problem caused by the severe background interferences and the tip characteristics of small size, being grayscale and indistinctive appearance patterns, tip detection in US images is challenging. METHODS: To achieve precise tip detection in US images against spatial indiscernibility, a novel multi-keyframe motion-aware framework called TipDet is proposed. It can identify tips based on their short-term spatial-temporal pattern and long-term motion pattern. In TipDet, first, an adaptive keyframe model (AKM) is proposed to decide whether a frame is informative to serve as a keyframe for long-term motion pattern learning. Second, candidate tip detection is conducted using a two-stream backbone (TSB) based on their short-term spatial-temporal pattern. Third, to further identify the true one in the candidate tips, a novel method for learning the long-term motion pattern of the tips is proposed based on the proposed optical-flow-aware multi-head cross-attention (OFA-MHCA). RESULTS: On the clinical human puncture dataset, which includes 4195 B-mode images, the experimental results show that the proposed TipDet can achieve precise tip detection against the spatial indiscernibility problem, achieving 78.7 % AP0.1:0.5 and 8.9 % improvement over the base detector at approximately 20 FPS. Moreover, a tip localization error of 1.3±0.6 % is achieved, exceeding the existing method. CONCLUSIONS: The proposed TipDet can facilitate a wider and easier application of US-guided interventional procedures by providing robust and precise needle tip localization. The codes and data are available at https://github.com/ResonWang/TipDet.

Coupling speckle noise suppression with image classification for deep-learning-aided ultrasound diagnosis.

Objective. During deep-learning-aided (DL-aided) ultrasound (US) diagnosis, US image classification is a foundational task. Due to the existence of serious speckle noise in US images, the performance of DL models may be degraded. Pre-denoising US images before their use in DL models is usually a logical choice. However, our investigation suggests that pre-speckle-denoising is not consistently advantageous. Furthermore, due to the decoupling of speckle denoising from the subsequent DL classification, investing intensive time in parameter tuning is inevitable to attain the optimal denoising parameters for various datasets and DL models. Pre-denoising will also add extra complexity to the classification task and make it no longer end-to-end.Approach. In this work, we propose a multi-scale high-frequency-based feature augmentation (MSHFFA) module that couples feature augmentation and speckle noise suppression with specific DL models, preserving an end-to-end fashion. In MSHFFA, the input US image is first decomposed to multi-scale low-frequency and high-frequency components (LFC and HFC) with discrete wavelet transform. Then, multi-scale augmentation maps are obtained by computing the correlation between LFC and HFC. Last, the original DL model features are augmented with multi-scale augmentation maps.Main results. On two public US datasets, all six renowned DL models exhibited enhanced F1-scores compared with their original versions (by 1.31%-8.17% on the POCUS dataset and 0.46%-3.89% on the BLU dataset) after using the MSHFFA module, with only approximately 1% increase in model parameter count.Significance. The proposed MSHFFA has broad applicability and commendable efficiency and thus can be used to enhance the performance of DL-aided US diagnosis. The codes are available athttps://github.com/ResonWang/MSHFFA.

Robust tip localization under continuous spatial and temporal constraints during 2D ultrasound-guided needle puncture.

PURPOSE: During ultrasound-guided (US-guided) needle puncture for minimally invasive procedures, automated needle tip localization can help clinicians capture small tips in US images easily and precisely, providing them with obvious tip indicators on the screen and bringing them more confidence during the procedures. However, automated needle tip localization in US images is challenging due to serious interferences arising from all kinds of echoes. METHODS: We propose a method that localizes needle tips under continuous spatial and temporal constraints in the real-time US frame stream. A temporal constraint is firstly acquired by detecting translational tip motion in motion-enhanced US images with a deep learning-based (DL-based) detector. A spatial constraint and candidate tip locations are obtained by detecting needle shafts and tips in the raw grayscale B-mode images with another DL-based detector. To provide continuous constraints, estimated tip velocity from acquired temporal constraint is used to predict tip locations in frames where no temporal or spatial constraint is detected. Finally, tip coordinates are precisely localized among candidate tips under the spatial and temporal constraints. RESULTS: Experimental results evaluated on 1121 US images from porcine organ punctures, and 895 images from human thyroid punctures demonstrate that the proposed method is effective and efficient, surpassing existing methods. On porcine organ data, a 97.2% recall rate and a 91.9% precision rate on tip detection and 0.88 ± 0.70 mm root-mean-square error (RMSE) on tip localization were achieved. On the human thyroid data, which was not involved in the training, 86.5% recall, 84.3% precision and 0.92 ± 0.78 mm RMSE were achieved separately. The running speed of 14.5 frames per second was achieved only using a CPU. CONCLUSION: The proposed method provides a more reliable solution for automated needle tip localization during US-guided needle puncture, being more robust to interferences. Fast running speed leads to its practicability in the real-time US stream.

[Study on the relationship between SNP of CD14 gene and allergic rhinitis in Xinjiang Uygurs and Hans].

OBJECTIVE: This study aimed to investigate the relationship between CD14 gene rs2569192(C/G), rs3138078 (--1359G/T) and allergic rhinitis (AR) in Xinjiang Uygur and Han populations as well as to determine characteristics of polymorphisms. METHOD: A total of 300 AR and 300 healthy controls subjects were included. The frequencies of genotypes and alleles were detected as well as the levels of tIgE in different genotypes were compared. RESULT: (1) The distribution of genotypes or alleles of CD14 gene rs2569192 (C/G), rs3138078 (--1359G/T) had no differences between the Xinjiang Uygurs and Hans (P > 0.05). The highest frequency of alleles was C, G. (2) The frequencies of genotypes and alleles were not different between the AR and control group in Uygur and Han (P > 0.05). The frequencies of genotypes and alleles of rs2569192 were different between the Uygur AR and Han AR group (P < 0.05). (3) The distribution of genotype frequencies and allele of rs 2569192 in the Xinjiang Uygur and Han population were quite different from Chinese Beijing Han populations, Japanese, European and African (P < 0.05). (4) The serum total IgE level in AR group was higher than that in healthy control group (P < 0.05). CONCLUSION: (1) rs2569192 (C/G), rs3138078 (--1359G/T) polymorphisms were not different between the Chinese Xinjiang Uygur and Han population. The major allele were both C and G. rs2569192 of CD14 in Xinjiang populations was different from that in the other populations. (2) No relationship between rs2569192, rs3138078 and AR was found. (3) The serum total IgE level in AR group was higher than that in healthy control group. No relationship between CD14 two SNP and serum total IgE level was found.