基于目标检测仔猪乳头计数及乳房形态评估方法

母猪乳头数量是生猪选育中重要参考指标之一,也是生猪繁殖表型数据的组成部分。成年母猪其腹部视频较难获取,且容易受污渍干扰,乳头自动点数实现难度较大,人工计数母猪乳头数工作量大、强度高、效率低、容易产生人为误差。鉴于猪仔从出生到成年乳头数量保持一致性,该研究提出了一种基于仔猪腹部视频的深度学习乳头计数及乳房形态评估方法。通过架设在仔猪初生护理平台上的相机拍摄仔猪腹部视频,根据清晰度筛选出细节清晰的帧序列图像集,经过数据预处理再使用改进Pignip-YOLOv5s目标检测网络对仔猪乳头进行自动计数。为提高计数准确率,帧序列图像集的乳头计数使用滑动窗口取众数得到最终计数结果。试验结果表明,改进的Pignip-YOLOv5s平均精度值(mean average precision, mAP)高达0.97,较YOLOv5原模型具备更高的鲁棒性。该研究方案在113段仔猪腹部视频数据集上测试得到仔猪乳头计数方法准确率达90.26%。同时该研究提出仔猪乳房形态评估参数乳头成对数、乳头间距,从而量化仔猪乳头形态表型特征,构建了母猪乳房外在形态指标,可为母猪选育和繁殖工作提供重要的参考依据。

产后乳房评估与产后乳房康复技能训练装置的研制

研制一款产后乳房评估与乳房康复技能训练的装置。方法 本装置由正常乳房、产后常见问题乳房、面板及背带组成,将不同设计的模块设置在面板上,借助两条背带穿戴于胸前。结果 使用该装置能有效地进行乳房评估,判断产后乳房常见问题,进行乳腺疏通规范化操作示教和疏通手法的反复练习,缓解乳汁淤积,预防急性乳腺炎。结论 教会产妇和护理人员针对产后生理性乳胀、乳汁淤积进行处理,对促进母乳喂养,促进婴幼儿健康成长具有积极作用,也为产后乳腺疏通按摩的标准化、规范化发展起到重要促进作用。

Preliminary results of acoustic radiation force impulses (ARFI) ultrasound imaging of solid suspicious breast lesions

Objective： The aim of our study was to make the qualitative and quantitative analysis to breast lesions using acoustic radiation force impulses （ARFI）, and assess the diagnostic value of ARFI for differentiation between benign and malignant solid breast masses, meanwhile evaluate the influences of ARFI with breast imaging reporting and data system （BI-RADS） of suspicious masses. Methods： Seventy-five women with 86 breast lesions underwent conventional breast ultra- sound examination. Then B-mode BI-RADS features and assessments were recorded and standard breast US supplemented by ARFI elastographic examination were repeated. The data were recorded and analyzed as following： area ratio of breast lesion, the shear-wave velocity, the ratio of the shear-wave velocity between lesions and surrounding normal tissues, and according to the elastographic data reconsidered the BI-RADS category, all the results have been correlated with pathological results and make statistical evaluations of ARFI for differentiation between benign and malignant solid breast masses. Meantime our study has correlated the adjusted BI-RADS category of suspicious breast lesions with the pathological results and made assessment. Results： Thirty-eight patients were malignant breast carcinoma （31 invasive ductal carcinoma, 5 intraductal carcinoma in situ, 2 medullary carcinoma, 2 invasive Iobular carcinoma）, 48 patients were benign breast lesions （23 fibroadenoma, 12 benign nodular hyperplasia, 5 phyllodes tumor, 6 adenosis, 2 intraductal papilloma）. Underwent conventional breast ultrasound exam, 42 cases were BI-RADS category 3, 23 cases were BI-RADS category 4. When adding elastographic data, 46 cases were BI-RADS category 3 and 20 cases were BI-RADS category 4. Compared with pathological results showed for both the specificity of BIRADS features and the area under ROC curve has risen. Virtual touch tissue imaging （VTI） and virtual touch tissue quantification （VTQ） data showed the area ratio （AR） between elastographic lesions area and B-mode lesions area, SWV （maximal shear-wave velocity of lesions）, R-SWV （shear-wave velocity ratio between lesions and surrounding normal tissues） in benign breast lesions were lower than those in malignant lesions which has statistical significance and the cut-off point were 1.1,4.65 m/s, 5.18 respectively. Conclusion： The ARFI elastography can provide the reliable qualitative and quantitative analysis about hardness of breast lesions, supply the new BI-RADS category features to suspicious breast masses and serve as an effective diagnostic tool for differentiation between benign and malignant solid masses.