OpenCASA: A new open-source and scalable tool for sperm quality analysis

Abstract

In the field of assisted reproductive techniques (ART), computer-assisted sperm analysis (CASA) systems have proved their utility and potential for assessing sperm quality, improving the prediction of the fertility potential of a seminal dose. Although most laboratories and scientific centers use commercial systems, in the recent years certain free and open-source alternatives have emerged that can reduce the costs that research groups have to face. However, these open-source alternatives cannot analyze sperm kinetic responses to different stimuli, such as chemotaxis, thermotaxis or rheotaxis. In addition, the programs released to date have not usually been designed to encourage the scalability and the continuity of software development. We have developed an open-source CASA software, called OpenCASA, which allows users to study three classical sperm quality parameters: motility, morphometry and membrane integrity (viability) and offers the possibility of analyzing the guided movement response of spermatozoa to different stimuli (useful for chemotaxis, thermotaxis or rheotaxis studies) or different motile cells such as bacteria, using a single software. This software has been released in a Version Control System at Github. This platform will allow researchers not only to download the software but also to be involved in and contribute to further developments. Additionally, a Google group has been created to allow the research community to interact and discuss OpenCASA. For validation of the OpenCASA software, we analysed different simulated sperm populations (for chemotaxis module) and evaluated 36 ejaculates obtained from 12 fertile rams using other sperm analysis systems (for motility, membrane integrity and morphology modules). The results were compared with those obtained by Open-CASA using the Pearson's correlation and Bland-Altman tests, obtaining a high level of correlation in all parameters and a good agreement between the different used methods and the OpenCASA. With this work, we propose an open-source project oriented to the development of a new software application for sperm quality analysis. This proposed software will use a minimally centralized infrastructure to allow the continued development of its modules by the research community.

Fig 1. OpenCASA overview diagram.

The software architecture was designed to facilitate the subsequent development of new features, so the code was separated in different packages depending on its functionality. Using these packages, the program allows user to carry out four different sperm analyses through the corresponding modules: Chemotaxis, Motility, Morphometry and Viability. In addition, a fifth module was implemented to generate simulations of chemotactically attracted sperm populations.

Fig 2. Definition of the instantaneous directionality angle.

ψ is the angle between the vector of the cell frame-to-frame displacement (${\overrightarrow{v}}_t=⟨P_t,P_{t+1}⟩$) and the gradient direction $\overrightarrow{\theta }$. In the example above, the gradient has been set to $\overrightarrow{\theta }=0°$.

Fig 3. Definition of two different options in order to analyse the chemotaxis phenomena.

It is important to count the number of instantaneous displacements ${\overrightarrow{v}}_t$ pointing in the chemotaxis gradient direction $\overrightarrow{\theta }$ and the number of those displacements not pointing to the gradient. $\psi =⟨\overrightarrow{v},\overrightarrow{\theta }⟩$ being the angle between the instantaneous displacement of a cell and the gradient direction; N⁺ is defined as count{ψ∈[−γ,+γ]}, where γ is a parameter defined by the user and represents the amplitude of the chemoattractant concentration gradient. The developed software allows users to choose two options to define which displacements not pointing to the gradient are taken into account. In option 1, N⁻ is defined as count{ψ∉[−γ,+γ]}, whereas in option 2 only displacements in the opposite direction of the gradient are considered (N⁻ = count{ψ∈[180°−γ,180°+γ]}). The images above show graphically which angles are taken into account for the sum N⁺(green color) and N⁻(red color), depending on the option specified by the user.

Fig 4. Two examples of the distribution of the instantaneous directionality angles ψ.

(a) Simulated sperm population without chemotaxis. (b) Simulated sperm population chemoattracted to θ = 0° (right). The ch-index provides information about the percentage of the angles ψ that point in the gradient direction with respect to the total number of angles taken into account (the total number of angles will depend on the option specified by the user). The parameters used to generate the simulation on the right were β = 1 and Responsiveness = 50%.

Fig 5. Determination of the O.R. threshold used to discriminate between chemotaxis and no chemotaxis in the bootstrapping method.

The histogram comes from 10000 O.R. ratios, each one calculated by the odds value of two disjointed subsets of trajectories randomly sampled over all detected trajectories in 100 control simulations. Each simulation consisted of a 500 frames length video (800x800 pixels each frame) containing 100 virtual cells randomly located at the beginning of the simulation. Each cell was defined as an ellipse (10x8 pixels size) and behaved following a persistent random walk equation with parameters D_rot = 0.1, v₀ = 3,β = 0, Reponsiveness = 0 and ψ₀ = 0°.

Fig 6. Membrane integrity module workflow.

The module receives an RGB image as input, splits the image into red and green channels, identifies viable and non-viable cells depending on the channel and finally merges all the results showing all the detected cells in the same image. The module identifies the viable cells in green and the non-viable cells in red.

Fig 7. Verification of the ch-index in a non-chemotaxis condition.

The ch-index provides information about the percentage of the angles ψ that point in the gradient direction with respect to the total of angles taken into account. In the case of a non- chemotaxis condition, a uniform distribution of the instantaneous directionality angles is expected, so defining γ = 30° and considering the angles ψ in the range [−30°,+30°] as chemotactical responses to the gradient, theoretically the percentage of those angles with respect to the total number of angles would be $\frac{60°}{360°}*100\approx 16.67\%$. Analysing the histogram, as expected, the ch-index was centred close to the theoretical value (17.90±0.46%) with some variation due to the sampling and the noise of the system (e.g. intersection of trajectories).