Detection of spermatogonial stem/progenitor cells in prepubertal mouse testis with deep learning

Kahveci, Burak; Önen, Selin; Akal, Fuat; Korkusuz, Petek

doi:10.1007/s10815-023-02784-1

Cited by 2 publications

(2 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Taking infertility as an example, which affects one in six couples worldwide [3,4], numerous deep learning models have been developed with the aim of improving clinical outcomes and optimizing the operational efficiency in in vitro fertilization (IVF) clinics [5][6][7][8]. Most of these models take images as input, for instance, to evaluate sperm motility, concentration, and morphology for selecting high-quality sperm for fertilization [9][10][11] or for diagnosing male infertility [12][13][14], to help identify and distinguish sperm and debris in testicular sperm samples [15,16], or to examine the quality of oocytes [17]. Models have also been developed to use embryo images or time-lapse videos to grade embryos [18,19] and to predict treatment outcomes such as implantation [20], pregnancy [21], and live birth [22][23][24].…”

Section: Introductionmentioning

confidence: 99%

“…Regardless of applications or the types of cells to analyze, the first technical step for deep learning models is often to visually identify and locate an object (oocyte [33,34], sperm [35][36][37][38][39], and embryo [20,[40][41][42]) in images. Different clinics, however, use different image acquisition conditions (e.g., microscope brands and models, imaging modes [43][44][45], magnifications [9,33], illumination intensity, and camera resolutions [13][14][15]39] etc. ), as evident in Table 1.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Testing the reproducibility and effectiveness of deep learning models among clinics: sperm detection as a pilot study

Wang,

Jin,

Jiang

et al. 2024

Preprint

View full text Add to dashboard Cite

Background: Deep learning has been increasingly investigated for assisting clinical in vitro fertilization (IVF). The first technical step in many tasks is to visually detect and locate sperm, oocytes, and embryos in images. For clinical deployment of such deep learning models, different clinics use different image acquisition hardware and different sample preprocessing protocols, raising the concern over whether the reported accuracy of a deep learning model by one clinic could be reproduced in another clinic. Here we aim to investigate the effect of each imaging factor on the reproducibility of object detection models, using sperm analysis as a pilot example. Methods: Ablation studies were performed using state-of-the-art models for detecting human sperm to quantitatively assess how model precision (false-positive detection) and recall (missed detection) were affected by imaging magnification, imaging mode, and sample preprocessing protocols. The results led to the hypothesis that the richness of image acquisition conditions in a training dataset deterministically affects model reproducibility. The hypothesis was tested by first enriching the training dataset with a wide range of imaging conditions, then validated through internal blind tests on new samples and external multi-center clinical validations. Results: Ablation experiments revealed that removing subsets of data from the training dataset significantly reduced model precision. Removing raw sample images from the training dataset caused the largest drop in model precision, whereas removing 20x images caused the largest drop in model recall. by incorporating different imaging and sample preprocessing conditions into a rich training dataset, the model achieved an intraclass correlation coefficient (ICC) of 0.97 (95% CI: 0.94-0.99) for precision, and an ICC of 0.97 (95% CI: 0.93-0.99) for recall. Multi-center clinical validation showed no significant differences in model precision or recall across different clinics and applications. Conclusions: The results validated the hypothesis that the richness of data in the training dataset is a key factor impacting model reproducibility. These findings highlight the importance of diversity in a training dataset for model evaluation and suggest that future deep learning models in andrology and reproductive medicine should incorporate comprehensive feature sets for enhanced reproducibility across clinics.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Testing the reproducibility and effectiveness of deep learning models among clinics: sperm detection as a pilot study

Wang,

Jin,

Jiang

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

Testing the generalizability and effectiveness of deep learning models among clinics: sperm detection as a pilot study

Wang,

Jin,

Jiang

et al. 2024

Reprod Biol Endocrinol

View full text Add to dashboard Cite

Background Deep learning has been increasingly investigated for assisting clinical in vitro fertilization (IVF). The first technical step in many tasks is to visually detect and locate sperm, oocytes, and embryos in images. For clinical deployment of such deep learning models, different clinics use different image acquisition hardware and different sample preprocessing protocols, raising the concern over whether the reported accuracy of a deep learning model by one clinic could be reproduced in another clinic. Here we aim to investigate the effect of each imaging factor on the generalizability of object detection models, using sperm analysis as a pilot example. Methods Ablation studies were performed using state-of-the-art models for detecting human sperm to quantitatively assess how model precision (false-positive detection) and recall (missed detection) were affected by imaging magnification, imaging mode, and sample preprocessing protocols. The results led to the hypothesis that the richness of image acquisition conditions in a training dataset deterministically affects model generalizability. The hypothesis was tested by first enriching the training dataset with a wide range of imaging conditions, then validated through internal blind tests on new samples and external multi-center clinical validations. Results Ablation experiments revealed that removing subsets of data from the training dataset significantly reduced model precision. Removing raw sample images from the training dataset caused the largest drop in model precision, whereas removing 20x images caused the largest drop in model recall. by incorporating different imaging and sample preprocessing conditions into a rich training dataset, the model achieved an intraclass correlation coefficient (ICC) of 0.97 (95% CI: 0.94-0.99) for precision, and an ICC of 0.97 (95% CI: 0.93-0.99) for recall. Multi-center clinical validation showed no significant differences in model precision or recall across different clinics and applications. Conclusions The results validated the hypothesis that the richness of data in the training dataset is a key factor impacting model generalizability. These findings highlight the importance of diversity in a training dataset for model evaluation and suggest that future deep learning models in andrology and reproductive medicine should incorporate comprehensive feature sets for enhanced generalizability across clinics.

show abstract

Detection of spermatogonial stem/progenitor cells in prepubertal mouse testis with deep learning

Cited by 2 publications

References 30 publications

Testing the reproducibility and effectiveness of deep learning models among clinics: sperm detection as a pilot study

Testing the reproducibility and effectiveness of deep learning models among clinics: sperm detection as a pilot study

Testing the generalizability and effectiveness of deep learning models among clinics: sperm detection as a pilot study

Contact Info

Product

Resources

About