DEGA-178-image-quantification #58
Conversation
| plt.figure(figsize=(10, 10)) | ||
| plt.imshow(stain_image_with_contrast, cmap='gray') | ||
| plt.title(f"Stain Image with Contrast Factor {contrast_factor}") | ||
| plt.show() |
There was a problem hiding this comment.
I saw a few plt.show() in this file. If it is standard QC results plot, it is better to write to a file. If this function is called multiple time, you may want to consider skipping the plt.show() with `plt.close() or add a bool argument to switch on/off displaying plot.
There was a problem hiding this comment.
During the optional contrast adjustment process, the displayed plots are only for checking whether the contrast level for the image is sufficient or not. And the data used to create the plots that have the stained image and the stain-positive cells overlaid on them is saved in a .csv file as well. Do you think it's required to save these plots separately? If you think it is better to save them too, I can add that line of code. Also, I have used plt.close() wherever needed
|
|
||
| return f"{base_path}/contrast_adjusted_stain_image.tif" | ||
|
|
||
| def process_inputs_for_quantification(technology_name, cell_boundary_file, image_file, transform_file, base_path, image_contrast_required=False, contrast_factor=5, intensity_threshold=100): |
There was a problem hiding this comment.
Some docstring below the function name can be very helpful here.
| cell_polygons["mean_intensity_EdU"] = intensities | ||
| cell_polygons["sum_intensity_EdU"] = sum_intensities | ||
|
|
||
| log_transformed_EdU_intensities = np.log1p(cell_polygons['sum_intensity_EdU']) |
There was a problem hiding this comment.
Why log1p is used here and not other normalization methods?
There was a problem hiding this comment.
I used log1p normalization because the image data I was working with had very low intensities for a specific stain, making the distribution highly skewed. I chose log1p to reduce the impact of extreme values and bring the data closer to a normal distribution. Do you think this is the right normalization method? If so, do you have any suggestions for a better normalization method?
|
|
||
| print("Filtered stain positive cells data (.csv) saved. Filtering done.") | ||
|
|
||
| def image_quantification(base_path, technology_name, image_contrast_required=False): |
There was a problem hiding this comment.
Do you have an example notebook or something that showcases this function? Would like to see an example. Thanks!
There was a problem hiding this comment.
Just sent you an example notebook and an example dataset!
No description provided.