شناسایی و شمارش سلولهای بنیادی تمایزیافته به بافت چربی از روی تصاویر میکروسکوپی با استفاده از تکنیکهای پردازش تصویر و بازشناسی الگو
Automatic Detection and Counting of Stem Cells Differentiated into Fat Tissue on Microscopic Images using Image Processing and Pattern Recognition Techniques
/لیلا حسنلو
: مهندسی برق و کامپیوتر
، ۱۳۹۷
، افشار
۹۳ص
چاپی - الکترونیکی
کارشناسی ارشد
مهندسی پزشکی گرایش بیوالکتریک
۱۳۹۷/۰۶/۱۹
تبریز
سلولهای بنیادی گروهی از سلولهای مستعد بوده که دارای توانایی تجدیدشوندگی و تمایز به انواع بافتهای استخوانی، چربی و غضروفی هستند .آنها نقش مهمی را در زمینه پزشکی احیاکننده ایفا میکنند که هدف آن بهبود و جایگزینی سلولها، بافتها و اندامهای بیمار از طریق پیوند سلولهای سالم و بافتها، بهویژه سلولهای بنیادی است .همچنین، علاقه و چالش در زمینه سلول بنیادی از این ناشی میشود که به بسیاری از مردم مبتلا به بیماری، معلولیت و نقص عضو، امید به زندگی میدهد .در حالت تمایز سلولهای بنیادی بهویژه تمایز به چربی که جهت انجام ترمیم بافت پوست استفاده میشود قطرات چربی معمولا در تعداد زیادی در داخل آنها تجمع مییابند .این قطرات چربی برای بررسی درصد تمایز به چربی در آزمایشهای مختلف توسط زیستشناسان سلولی بهصورت دستی شمرده میشوند .همچنین، شناسایی سلولهای تمایزیافته به چربی از اهمیت بالایی برخوردار است .روش متعارف برای مشاهده و شمارش قطرات چربی و تشخیص سلولهای تمایزیافته، رنگآمیزی با Oil red است که سبب آسیب دیدن سلولها میشود .علاوه بر آن، شمارش قطرات چربی بهصورت دستی میتواند یک کار سخت و دشوار باشد .همچنین، پرهزینه بودن این روشها و متفاوت بودن تفسیرهای هر متخصص، از معایب روش شمارش دستی است
Stem cells are a group of competent cells capable of self-renewal and differentiation into osteogenic, chondrogenic, and adipogenic lineages. They play an important role in the field of regenerative medicine, which aims to improve and replace the cells, tissues and organs of the patient through the transplantation of healthy cells and tissues, especially stem cells. Also, the interest and challenge in the stem cells context are due to the fact that gives life expectancy to many people with a disease, disability and organ failure. In the case of stem cells differentiation especially the adipocyte differentiation that used for skin tissue repair lipid droplets usually accumulate in a large number inside them. Also, identification of differentiated stem cells to adipose is of great importance. These lipid droplets are manually counted by biologists to determine the percentage of adipocyte differentiation in different experiments. The conventional method for observing and counting them and detection of differentiated cells is staining with Oil Red, which can damage the cells. In addition, counting manually can be a difficult task. Also, the cost of these methods and the different interpretations of each specialist are the disadvantages of manual counting. Research on the processing of intracellular images has begun while the optimal algorithms and techniques in these images are much less than other medical image processing; because these images have a high magnification and low resolution. Therefore, there is no intelligent system for auto-particle counting in intracellular images. In this thesis, after the cultivation of mesenchymal stem cells and their differentiation into adipose tissue, their intracellular images were prepared. Then, a deep learning-based method for counting lipid droplets produced in the differentiation and identification of differentiated cells is presented. In the proposed method, microscopic images are divided into square images; then, lipid droplets are labeled as the single point. The proposed network is a fully convolutional regression network that processes an image with a receptive field on the image, and finally, in addition to predicting the number of lipid droplets, produces a count map that is useful for determining the adipocyte differentiation rate and differentiated cells in different samples by cell biologists. The average accuracy of the count is 94 , which is higher than state-of-the-art methods. By generating a count map, differentiated mesenchymal stem cells can be detected without oil-red coloration and prevent cell damage by staining. The contribution of this thesis is that the deep learning algorithm has been used for the first time in the field of intracellular image processing
Automatic Detection and Counting of Stem Cells Differentiated into Fat Tissue on Microscopic Images using Image Processing and Pattern Recognition Techniques