AI-Assisted Support System in Sperm Analysis

How AI Image Processing Can Help Sperm Analysis

Turning Microscope Images Into Measurable Data

Semen analysis is a basic lab test for male fertility. It is also called a spermogram. It checks the number of sperm, how they move and their shape.

Today, AI image processing can find cells in microscope images on its own. It can also track how they move and turn this into numbers.

Our centre has built a software prototype as a research and development project. It aims to:

• find sperm cells in microscope images with AI
• track each cell from one frame to the next
• measure basic movement features

The goal is not to replace lab evaluation. The goal is to build a more measurable, repeatable and traceable standard for sperm analysis.

Why Is AI Image Processing Useful in Sperm Analysis?

Sperm cells are small and keep moving under the microscope. They may come close to each other, overlap or leave the image. So finding many cells at once and tracking each one takes skill.

AI image processing can help find sperm-like shapes in microscope videos. It can also measure how these cells move over time.

Our prototype checks each cell by how it looks, including its:

• size
• shape
• roundness
• oval form
• contrast with the background

The software matches each cell between video frames. Then it draws a separate movement path for each one.

This way, movement is not only looked at by eye. It becomes data that can be measured and compared.

How AI Finds Sperm Cells

First, the microscope image is prepared. It is turned to greyscale. Noise is reduced, and contrast is raised so cells stand out from the background.

Next, the software finds small shapes that look like sperm heads. It checks each one for area, shape, size and contrast.

Grid lines, air bubbles, cell debris and other marks can cause false results. So the software uses extra filters to remove shapes that are not sperm.

At this stage, the software does not treat what it finds as sperm for certain. It treats them as sperm cell candidates that meet its rules.

How AI Tracks Sperm Movement

After finding the candidates, the software tries to follow each one through the video.

It compares where a candidate is in one frame with the objects in the next frame. The object that best fits by distance and direction is taken as the same cell.

This creates a movement trace for each candidate over time. The tracking can help measure:

• total distance moved
• direction of movement
• how steady the movement is
• how straight the cell moves
• changes in direction
• distance moved in a set time

Recording the movement of many cells one by one can support a more detailed study of sperm motility.

Repeatability

The software uses fixed rules. So the same image can be analysed the same way at different times. This makes it possible to compare results and measure how the system improves.

Tracking Many Cells at Once

The human eye can find it hard to follow many moving sperm cells one by one. AI systems can track many cells at the same time and draw a path for each one.

Movement (Kinematic) Analysis

AI image processing helps measure the speed and paths of sperm cells. Measures such as VCL (curved-line speed) and VSL (straight-line speed) show movement in more detail.

How It Could Support the Lab

Once scientific and clinical testing is complete, such systems may help with:

• pre-checking images
• marking sperm cells
• drawing movement paths
• preparing measurements
• supporting the specialist’s review

The aim is not to replace the lab specialist. It is to give the specialist organised, measurable data.

Scientific Validation

Before use in sperm analysis, the results must be compared with expert reviews and accepted lab methods.

Validation looks at:

• how often sperm cells are found correctly
• how often other shapes are wrongly counted as sperm
• the difference between automatic and expert counts
• how well cells are tracked without breaks
• how accurate the movement groups are
• how reliable the movement measures are
• results across different samples, microscopes, cameras and image quality

Both where the system works well and where it makes mistakes must be shown openly.

Does AI Replace the Specialist?

No. A single AI result cannot show whether a person is fertile or not.

A specialist doctor must review semen analysis together with your medical history, physical exam, hormone and genetic tests and other findings.

Sperm values can change from day to day. So the test should be repeated when needed. AI systems do not replace expert decisions. They give numbers and images that support the review.

Our prototype is a research and development project. It finds sperm cell candidates in microscope images, tracks their movement and measures basic movement features.

Its main possible benefits are tracking many cells at once, recording movement paths, calculating movement measures and supporting the specialist with data.

But the system has not been clinically validated in its current form. Its accuracy, reliability and possible uses can only be known after controlled scientific studies that compare it with experts.

Important information: This AI image processing software is a research prototype in development. It is not a medical device or a diagnostic tool. It does not diagnose fertility or infertility. It does not recommend treatment. It must not be used to make health decisions. It does not replace expert lab evaluation. Semen analysis must be done in proper lab conditions by trained staff, and the results must be read by a specialist doctor.