gaussquality.gaussquality_io
Gaussian mixture model fitting for greyscale images: IO
Created on Mon Jan 25 14:50:04 2021
@author: elainehoml
View Source
# -*- coding: utf-8 -*- """ Gaussian mixture model fitting for greyscale images: IO Created on Mon Jan 25 14:50:04 2021 @author: elainehoml """ import os import json import pandas as pd import matplotlib.pyplot as plt import skimage.io import datetime def get_img_list(img_dir): """ Gets list of .tiff images in an image sequence folder. Parameters ---------- img_dir : str, path-like Path to folder containing single images. Returns ------- list List of image filenames in the folder in ascending order. Only works on .tiff images. """ img_list_unsorted = [] for f in os.listdir(img_dir): if f.lower().endswith((".tif", ".tiff")): img_list_unsorted.append(os.path.join(img_dir, f)) return sorted(img_list_unsorted) def get_img_filepath(img_dir, index): """ Gets filepath for a specific image in a sequence. Parameters ---------- img_dir : str, path-like Path to folder containing single images. index : int Index of image Returns ------- str, path-like Image filepath for index image in sequence """ img_list = get_img_list(img_dir) return img_list[index] def get_nslices(img_dir): """ Gets number of slices in an image sequence folder. Parameters ---------- img_dir : str, path-like Path to folder containing single images. Returns ------- int Number of images in the sequence. Must ensure that no other files exist in the folder besides the images. """ return len(get_img_list(img_dir)) def mask_img(img, mask_percentage): """ Applies a mask in x-y plane to only consider the central `mask-percentage` percentage of the image. Parameters ---------- img : array-like 2-D array containing image grey values. mask_percentage : float Percentage of the image to consider, as a rectangle centred on `img`. Ranges from 0-100. Returns ------- masked_img : array-like Central `mask-percentage` of the 2D image `img` """ if mask_percentage < 0 or mask_percentage > 100: raise ValueError("`mask_percentage` must be a percentage between \ 0 and 100") mask_width = int(img.shape[0] * mask_percentage/100) mask_height = int(img.shape[1] * mask_percentage/100) centre_x, centre_y = (int(img.shape[0]/2), int(img.shape[1]/2)) x_bounds = (int(centre_x - mask_width/2), int(centre_x + mask_width/2)) y_bounds = (int(centre_y - mask_height/2), int(centre_y + mask_height/2)) masked_img = img[x_bounds[0]:x_bounds[1], y_bounds[0]:y_bounds[1]] return masked_img def load_img(img_filepath, show_image=False, mask_percentage=100., vmin=None, vmax=None): """ Loads image from `img_filepath` and applies a mask with percentage `mask_percentage`. Parameters ---------- img_filepath : str, path-like Filepath to the image to import. show_image : bool, optional If True, display the image. The default is False. mask_percentage : float, optional Percentage of the image to import, as a rectangle centred in the x-y plane. The default is 100. v_min : float, optional, default None Minimum grey value to plot v_max : float, optional, default None Maximum grey value to plot Returns ------- img : array-like 2-D array representing masked image. """ img = skimage.io.imread(img_filepath) masked_img = mask_img(img, mask_percentage) if show_image is True: plt.imshow(masked_img, cmap="gray", vmin=vmin, vmax=vmax) plt.title("{}\n{}".format(os.path.split(img_filepath)[-1], masked_img.shape)) return masked_img def save_GMM_single_results(fitted_results, save_dir, prefix, SNR=None, CNR=None): """ Saves average fitted Gaussian properties across the stack. Parameters ---------- fitted_results : list List containing fitted Gaussian properties `mu`, `sigma` and `phi` averaged across the stack. save_dir : str, path-like Directory to savepath. prefix : str Prefix to filename SNR : dict, optional. SNR. Default is None. CNR : dict, optional. CNR. Default is None. Returns ------- None. """ save_single_filename = os.path.join(save_dir, prefix + "_GMM_results.json") # unpack results mu_mean, sigma_mean, phi_mean = fitted_results # create dict dict_to_write = {} for material in range(len(mu_mean)): dict_to_write["material_" + str(material)] = [] dict_to_write["material_" + str(material)].append({ "mu_mean": mu_mean[material], "sigma_mean": sigma_mean[material], "phi_mean": phi_mean[material] }) if (SNR is not None) and (CNR is not None): dict_to_write["SNR"] = SNR dict_to_write["CNR"] = CNR with open(save_single_filename, "w") as outfile: json.dump(dict_to_write, outfile, indent=4) def save_GMM_slice_results(iter_results, save_dir, prefix): """ Saves fitted Gaussian properties for each individual 2-D image considered from a 3-D image sequence Parameters ---------- iter_results : list List of fitted `mu`, `sigma` and `phi` Gaussian properties. save_dir : str, path-like Directory to save results. prefix : str Prefix to save results filename. Returns ------- None. """ parameters = ["mu", "sigma", "phi"] for parameter in range(len(parameters)): save_filename = os.path.join(save_dir, "{}_{}_GMM_slice_results.csv".format(prefix, parameters[parameter])) # pd.DataFrame(iter_results[parameter]).to_json(save_filename, indent=4) pd.DataFrame(iter_results[parameter]).transpose().to_csv(save_filename) def save_SNR_CNR_stack(SNRs, CNRs, save_dir, prefix): # deprecated dict_to_write = {"SNR": SNRs, "CNR": CNRs} snr_cnr_outfile = os.path.join( save_dir, "{}_snr_cnr.json".format(prefix)) with open(snr_cnr_outfile, "w") as outfile: json.dump(dict_to_write, outfile, indent=4) print("SNR and CNR saved as {}".format(snr_cnr_outfile))
View Source
def get_img_list(img_dir): """ Gets list of .tiff images in an image sequence folder. Parameters ---------- img_dir : str, path-like Path to folder containing single images. Returns ------- list List of image filenames in the folder in ascending order. Only works on .tiff images. """ img_list_unsorted = [] for f in os.listdir(img_dir): if f.lower().endswith((".tif", ".tiff")): img_list_unsorted.append(os.path.join(img_dir, f)) return sorted(img_list_unsorted)
Gets list of .tiff images in an image sequence folder.
Parameters
- img_dir (str, path-like): Path to folder containing single images.
Returns
- list: List of image filenames in the folder in ascending order. Only works on .tiff images.
View Source
def get_img_filepath(img_dir, index): """ Gets filepath for a specific image in a sequence. Parameters ---------- img_dir : str, path-like Path to folder containing single images. index : int Index of image Returns ------- str, path-like Image filepath for index image in sequence """ img_list = get_img_list(img_dir) return img_list[index]
Gets filepath for a specific image in a sequence.
Parameters
- img_dir (str, path-like): Path to folder containing single images.
- index (int): Index of image
Returns
- str, path-like: Image filepath for index image in sequence
View Source
def get_nslices(img_dir): """ Gets number of slices in an image sequence folder. Parameters ---------- img_dir : str, path-like Path to folder containing single images. Returns ------- int Number of images in the sequence. Must ensure that no other files exist in the folder besides the images. """ return len(get_img_list(img_dir))
Gets number of slices in an image sequence folder.
Parameters
- img_dir (str, path-like): Path to folder containing single images.
Returns
- int: Number of images in the sequence. Must ensure that no other files exist in the folder besides the images.
View Source
def mask_img(img, mask_percentage): """ Applies a mask in x-y plane to only consider the central `mask-percentage` percentage of the image. Parameters ---------- img : array-like 2-D array containing image grey values. mask_percentage : float Percentage of the image to consider, as a rectangle centred on `img`. Ranges from 0-100. Returns ------- masked_img : array-like Central `mask-percentage` of the 2D image `img` """ if mask_percentage < 0 or mask_percentage > 100: raise ValueError("`mask_percentage` must be a percentage between \ 0 and 100") mask_width = int(img.shape[0] * mask_percentage/100) mask_height = int(img.shape[1] * mask_percentage/100) centre_x, centre_y = (int(img.shape[0]/2), int(img.shape[1]/2)) x_bounds = (int(centre_x - mask_width/2), int(centre_x + mask_width/2)) y_bounds = (int(centre_y - mask_height/2), int(centre_y + mask_height/2)) masked_img = img[x_bounds[0]:x_bounds[1], y_bounds[0]:y_bounds[1]] return masked_img
Applies a mask in x-y plane to only consider the central mask-percentage
percentage of the image.
Parameters
- img (array-like): 2-D array containing image grey values.
- mask_percentage (float):
Percentage of the image to consider, as a rectangle centred on
img. Ranges from 0-100.
Returns
- masked_img (array-like):
Central
mask-percentageof the 2D imageimg
View Source
def load_img(img_filepath, show_image=False, mask_percentage=100., vmin=None, vmax=None): """ Loads image from `img_filepath` and applies a mask with percentage `mask_percentage`. Parameters ---------- img_filepath : str, path-like Filepath to the image to import. show_image : bool, optional If True, display the image. The default is False. mask_percentage : float, optional Percentage of the image to import, as a rectangle centred in the x-y plane. The default is 100. v_min : float, optional, default None Minimum grey value to plot v_max : float, optional, default None Maximum grey value to plot Returns ------- img : array-like 2-D array representing masked image. """ img = skimage.io.imread(img_filepath) masked_img = mask_img(img, mask_percentage) if show_image is True: plt.imshow(masked_img, cmap="gray", vmin=vmin, vmax=vmax) plt.title("{}\n{}".format(os.path.split(img_filepath)[-1], masked_img.shape)) return masked_img
Loads image from img_filepath and applies a mask with percentage
mask_percentage.
Parameters
- img_filepath (str, path-like): Filepath to the image to import.
- show_image (bool, optional): If True, display the image. The default is False.
- mask_percentage (float, optional): Percentage of the image to import, as a rectangle centred in the x-y plane. The default is 100.
- v_min (float, optional, default None): Minimum grey value to plot
- v_max (float, optional, default None): Maximum grey value to plot
Returns
- img (array-like): 2-D array representing masked image.
View Source
def save_GMM_single_results(fitted_results, save_dir, prefix, SNR=None, CNR=None): """ Saves average fitted Gaussian properties across the stack. Parameters ---------- fitted_results : list List containing fitted Gaussian properties `mu`, `sigma` and `phi` averaged across the stack. save_dir : str, path-like Directory to savepath. prefix : str Prefix to filename SNR : dict, optional. SNR. Default is None. CNR : dict, optional. CNR. Default is None. Returns ------- None. """ save_single_filename = os.path.join(save_dir, prefix + "_GMM_results.json") # unpack results mu_mean, sigma_mean, phi_mean = fitted_results # create dict dict_to_write = {} for material in range(len(mu_mean)): dict_to_write["material_" + str(material)] = [] dict_to_write["material_" + str(material)].append({ "mu_mean": mu_mean[material], "sigma_mean": sigma_mean[material], "phi_mean": phi_mean[material] }) if (SNR is not None) and (CNR is not None): dict_to_write["SNR"] = SNR dict_to_write["CNR"] = CNR with open(save_single_filename, "w") as outfile: json.dump(dict_to_write, outfile, indent=4)
Saves average fitted Gaussian properties across the stack.
Parameters
- fitted_results (list):
List containing fitted Gaussian properties
mu,sigmaandphiaveraged across the stack. - save_dir (str, path-like): Directory to savepath.
- prefix (str): Prefix to filename
- SNR (dict, optional.): SNR. Default is None.
- CNR (dict, optional.): CNR. Default is None.
Returns
- None.
View Source
def save_GMM_slice_results(iter_results, save_dir, prefix): """ Saves fitted Gaussian properties for each individual 2-D image considered from a 3-D image sequence Parameters ---------- iter_results : list List of fitted `mu`, `sigma` and `phi` Gaussian properties. save_dir : str, path-like Directory to save results. prefix : str Prefix to save results filename. Returns ------- None. """ parameters = ["mu", "sigma", "phi"] for parameter in range(len(parameters)): save_filename = os.path.join(save_dir, "{}_{}_GMM_slice_results.csv".format(prefix, parameters[parameter])) # pd.DataFrame(iter_results[parameter]).to_json(save_filename, indent=4) pd.DataFrame(iter_results[parameter]).transpose().to_csv(save_filename)
Saves fitted Gaussian properties for each individual 2-D image considered from a 3-D image sequence
Parameters
- iter_results (list):
List of fitted
mu,sigmaandphiGaussian properties. - save_dir (str, path-like): Directory to save results.
- prefix (str): Prefix to save results filename.
Returns
- None.
View Source
def save_SNR_CNR_stack(SNRs, CNRs, save_dir, prefix): # deprecated dict_to_write = {"SNR": SNRs, "CNR": CNRs} snr_cnr_outfile = os.path.join( save_dir, "{}_snr_cnr.json".format(prefix)) with open(snr_cnr_outfile, "w") as outfile: json.dump(dict_to_write, outfile, indent=4) print("SNR and CNR saved as {}".format(snr_cnr_outfile))