Ovarian Follicle Detection

Abstract

Ovarian follicle detection is an effective tool used in infertility treatmen t. Here a new algorithm of modified fuzzy logic is use d for ovarian follicle recognition from a series of ultrasound images. Prolo nged time and variable resu lts by the various clinicians make the detection inconvenient for the patients , thus making automat ic and fast detection of follicular monitoring necessary. Noise, detecting multiple follicles that are very close to each other as one follicle region withou t finding the boundar y of indiv idual follicles, and not being fast enough to be used in real -time c linical practice are the limiting factors which results in low performance of computerized detection .

To overcome these limitations, we handle noise by singu lar value decompositi on -based image compression f ollowed by an anisotropic diffusion scheme for multiplicative speckle, and detect follicles by performing different segmentation techniques like modified fuzzy and contour segmentation . The classification of the follicles is performe d usi ng back propagatio n neural network and the detected result is sent to the patient via email.

Get quality help now

Doctor Jennifer

Verified writer

Proficient in: Health

5 (893)

“ Thank you so much for accepting my assignment the night before it was due. I look forward to working with you moving forward ”

+84 relevant experts are online

Hire writer

INTRODUCTION

Ultrasonography is a non -invasive imaging m odality of obstetrics and gynaecology [1]. In repr oductive medicine, transvaginal ultrasound examination is primarily used for monitoring follicular growth during ovarian stimulation and for estimating ovari an res erve [2]. The number and size of ovarian fo llicles, number of antral follicles (follicles tha t are 2-8 mm in average diameter) [3], and growth rate of dominant follicles (follicles that are larger than 10 mm in average diameter) [4] are the primary e ndpo ints of measurement for ova rian follicle moni toring [2].

Ovarian follicular monitoring is essen tial for guiding the amount and duration of medications for ovarian  timulation, ovulation induction and intrauterine insemination, controlled ovarian hypers timul ation for oocyte (egg) ret rieval, in vitro fertilization (IVF) and fresh embryo transfer, egg donation cycles, and for women who electively freeze their eggs or embryos for future use. Performing ultrasound scans every 2-3 days is necessary to optimiz e ova rian response, the yield of eggs, and the ul timate success of an IVF cycle. Since measuring fo llicles is done manually and requires multiple visits, it becomes tremendously inconvenient given the number of examinations that need to be done at fertility centres and hospitals [ 5].

Moreover, ultrasound images analysed by different technicians or medic al experts can lead to inconsistent results and interpretations [6]. Hence, automated follicular monitoring has the great potential to maximize pregnancy s uccess of IVF treatment on a large scale. In medica l image processing, first images are pre-processed and enhanced, then features are extracted and selected, and finally, images are classified and segmented [7].

Here we use modified fuzzy and contour segmentation for segmentation purpose to get more accuracy with minimum processing time . The classification of the follicles is performe d using back propagation neural netw ork an d the detected result is sent to the patient via email .

METHODS

The ultrasound images of follicles were given as input to the follicle detection algorithm for de tecting the follicles in the inputted image without requiring the presence of a medical expert .

Folli cle Detection Algorithm

• Initially the image given as input is read .
• Color conversion to grayscale image from the rgb image is done by using the fun ction rgb2 gray in MATLAB R2014 a.
• Median filter is used to reduce noise in an image by preserving useful detail s in it and by removing the unwanted details from it. Histogram equalization is used to en hance the contrast of the i mage and the image is then convert ed to binary by thresholding .
• Modified fuzzy segmentation is used to incre ase the accurac y of segmen tation.
• Active contour anal ysis defines a separate boundary or curvature f or the regions of target object for segm entation. Thus , the active co ntour segmentation is used for th eseparation of pixels o f interest for different image processin g.
• GLCM feature ext raction gives a measure of the varia tion in intensi ty at the pixel of interest. It considers the relation between two pixel s at a time, ca lled t he reference and the neighboring pixel.
• Finally classification of image s is done using back propagation neural network and the report is sent to the user via email .

RESULTS

The proposed algorithm successfully detects the follicles and outperform the conventional methods.

CONCLUSIONS

The proposed system detects the follicles within few seconds. In the proposed algorithm modified fuzzy segmentation is used and assification of the follicles is performe d usi ng back propagatio n neural network which makes th e detection more time efficient an d the detected result is sent t o the patient via email .

ACKNOWLEDGEMENT

For the successful completion of th is project , I thank my guide Nithin Joe Assistant Professor in Department of Electronics and Communicat ion Engineering , NCERC, Pampady , India for providing valuable suggestions during the project. I thank all my guides, family members and friends who helped me directly or indirectly during this work.

REFERENCES

1. Deutchman, Mark E., and Ricardo Hahn. "Obstetricultrasonography." Primary Care: Clinics in Office Practice 24.2 (1997): 407 -431.
2. Lebbi, I., and R. Ben Temime. "The significanc e of monitoring folliculogenesis." IVF Lite 2.1 (2015): 6.
3. Depmann, Martine, et al. "Fluctuations in anti Mllerian hormone levels throughout the menstrual cycle parallel fluctuations in the antral follicle count: a cohort study." Acta obstetricia et g ynecologica Scandinavica 95.7 (2016): 820 -828.
4. Hu, Xiaokun, et al. "New Perspectives on Criteria for the Determination of HCG Trigger Timing in GnRH Antagonist Cycles." Medicine 95.20 (2016).
5. Potonik, Boidar, Boris Cigale, and Damjan Zazula. "Com puterized detection and recognition of follicles in ovarian ultrasound images: a review." Medical & biological engineering & computing 50.12 (2012): 1201 - 1212.
6. Ata, Baris, and Togas Tulandi , "Ultrasound automated volume calculation in reproduction and in pregnancy." Fertility and sterility 95.7 (2011): 2163 -2170.
7. Chatap, Niranjan J., and Ashish Kr Shrivastava. "A Survey on Various Classification Techniques for Medical Image Data." International Journal of Computer Applications 97.15 (2014).
8. Rose T. Faghih , Aaron K. Styer, Emery N. Brown Automated Ovarian Follicular Monitoring: A Novel Real -Time Approach IEEE (2017 ) 978 -1-5090 -2809 - 2/17 .
9. Rabiu, I. O., A. D. Usman, and A. M. S. Tekanyi. "A Review on Computer Assisted Follicle Detection Techniques and Polycystic Ovarian Syndrome (PCOS) Diagnostic Systems ." International Journal of Computer Trends and Technology (IJCTT) 1.6 (2012).
10. Hiremath, P. S., and Jyothi R. Tegnoor. "Automated ovarian classification in digital ultrasound images." International Journal of Biomedical Engineering and Technology 11.1 (2013): 46 -65.