screenCrop.py

'''
Created on Sep 5, 2017

Loads embryo images generated by CV1000 and crops/orients and saves separate embryos based on DIC (C3) central image
'''

date = '20180808T122436'
'''directly edit parameters'''
loadFolder = 'Z:/'
folderIn = loadFolder + 'CV1000/' + date  # Input folder
trackingFile = 'Z:/Experiment_tracking_sheets/EMBD_fileNames_Tracking_Sheet.csv'
aspectRatioFile = 'Z:/cropped/aspects.csv'
z = 18  # number of z planes
nT = 31  # number of time points
corrDrift = True
removeBG = True
attCorrect = True
apRotate = True
nWells = 1  # number of wells (14)
pointVisits = 1  # number of point visits (4)

'''read parameters from config file parameters.py'''
import parameters
if parameters.use_config:
    print("using parameters defined in the CONFIGURATION file (parameters.py)")
    date = parameters.date
    loadFolder = parameters.loadFolder
    folderIn = parameters.folderIn
    trackingFile = parameters.trackingFile
    aspectRatioFile = parameters.aspectRatioFile
    z = parameters.z
    nT = parameters.nT
    corrDrift = parameters.corrDrift
    removeBG = parameters.removeBG
    attCorrect = parameters.attCorrect
    apRotate = parameters.apRotate
    nWells = parameters.nWells
    pointVisits = parameters.pointVisits
else:
    print("using parameters defined in the SOURCE CODE")

import glob, csv, cv2, os, shutil
import numpy as np
from findEmbryo import showIm
from myFunc import clearFolder
from cropAPI import cropEmbs
from tkinter import messagebox as tkMessageBox

debug = False  # use to debug the program (line 101)


def getConditions(date, fileName):
    ''' loads RNAi strains for a specified date from a csv file '''
    global RNAi, strains
    #     csvFile = csv.reader(open(fileName, 'rb'), delimiter=',')
    csvFile = csv.reader(open(fileName, 'rU'), delimiter=',')  # universal
    fileData = []
    for row in csvFile:
        fileData.append(row[1:-1])
    myDate = [s for s in fileData if s[0] == date]
    myDate = sorted(myDate, key=lambda well: well[2])
    RNAi = [s[3] for s in myDate]
    strains = [s[4] for s in myDate]
    return


def loadImages(folder, well, j):
    '''
    loads images from a folder and splits them in separate point visits
    
    Parameters:
    folder : folder to read images from
    
    Return:
    allImgs: list of 4 different point visits with 3 channels in each. images are numpy arrays.
    '''
    imc1, imc2, imc3 = [], [], []
    folderNames = glob.glob(folder + '/Well{0:0>3}/*F{1:0>3}*C1.tif'.format(well, j))
    folderNames.sort()
    for fileName in folderNames:
        imc1.append(cv2.imread(fileName, -1))

    folderNames = glob.glob(folder + '/Well{0:0>3}/*F{1:0>3}*C2.tif'.format(well, j))
    folderNames.sort()
    for fileName in folderNames:
        imc2.append(cv2.imread(fileName, -1))

    folderNames = glob.glob(folder + '/Well{0:0>3}/*F{1:0>3}*C3.tif'.format(well, j))
    folderNames.sort()
    for fileName in folderNames:
        imc3.append(cv2.imread(fileName, -1))

    if len(imc1) > 0:
        imc1 = np.reshape(imc1, (nT, z, imc1[0].shape[-2], imc1[0].shape[-1]))
        imc2 = np.reshape(imc2, (nT, z, imc2[0].shape[-2], imc2[0].shape[-1]))
        imc3 = np.reshape(imc3, (nT, z, imc3[0].shape[-2], imc3[0].shape[-1]))
        allImgs = np.stack((imc1, imc2, imc3), axis=2)
        return allImgs
    else:
        return np.array((imc1, imc2, imc3))


def getAllEmb(folder):
    '''
    Finds and saves all embryos
    
    ParametersL
    folder: folder to load embryos
    
    Return:
    None
    '''
    global RNAi, strains

    print('STARTED!!!')
    getConditions(date, trackingFile)
    totalEmb = 0
    embs = []

    for well in range(nWells):
        for j in range(pointVisits):
            if debug:
                well, j = 2, 1  # well, field
            imgs = loadImages(folder, well + 1, j + 1)
            if imgs.size > 0:
                print('loaded well {0} point {1}'.format(well + 1, j + 1))
                if strains[well] == 'MS':
                    featureList = [201, 201, None]
                else:
                    featureList = [41, 41, None]
                es, rs = cropEmbs(imgs, 2, corrDrift, attCorrect, 0.1, removeBG, featureList, 0.315, EmbdScreen=True)
                embs.append([well, j, es, rs])
                totalEmb += len(embs[-1]) - 2
            else:
                print('no images well {0} point {1}'.format(well + 1, j + 1))
            del imgs

    print('Done analyzing, ready to save!')
    checks = np.array([checkEmbDebris(e[2 * 3 * z + 3 * z // 2 + 2]) for well, j, es, rs in embs for e in es])
    totalEmb = checks.size
    uniqeRNAi = np.array(list(set(RNAi)))
    embN = np.ones([len(uniqeRNAi), 2])
    xembN = np.ones([len(uniqeRNAi), 2])
    if not debug:
        for well, j, es, rs in embs:
            if strains[well] != 'MS':
                k = 0
            else:
                k = 1
            for l in range(len(es)):
                e = es[l]
                r = rs[l]
                print('{0} embryos left, saving...'.format(totalEmb))
                i = checks.size - totalEmb
                if checks[i] == 1:
                    saveEmb(e, j, int(embN[np.where(uniqeRNAi == RNAi[well])[0], k]), well, checks[i], r)
                    embN[np.where(uniqeRNAi == RNAi[well])[0], k] += 1
                elif checks[i] == 2:
                    saveEmb(e, j, int(xembN[np.where(uniqeRNAi == RNAi[well])[0], k]), well, checks[i], r)
                    xembN[np.where(uniqeRNAi == RNAi[well])[0], k] += 1
                totalEmb -= 1


def checkEmbDebris(im):
    ''' lets user debug supplied image, and returns 1 for yes (save) and 0 for not and 2 for special case'''
    code = showIm(im)
    if code == ord('d') or code == ord('D'):
        result = tkMessageBox.askquestion("Delete", "Are You Sure?", icon='warning')
        if result == 'yes':
            return 0
        else:
            return checkEmbDebris(im)
    elif code == ord('x') or code == ord('X'):
        return 2
    else:
        return 1


def saveEmb(imgs, point, i, well, check, r):
    '''
    Saves embryo images according to a certain pattern
    
    Parameters:
    imgs: numpy array with images in czt order
    f: point visit number
    i: embryo number
    r: aspect ratio
    '''

    imgs = np.reshape(imgs, (nT, z, 3, imgs.shape[-2], imgs.shape[-1]))
    im1 = imgs[:, :, 0]
    im2 = imgs[:, :, 1]
    im3 = imgs[:, :, 2]
    strain = strains[well]
    ri = RNAi[well]
    if ri != 'EMBD0000':
        folderOut = loadFolder + 'cropped/{0}/{1}/'.format(ri, strain)  # outputFolder
    else:
        folderOut = loadFolder + 'cropped/EMBD0000/{0}/{1}/'.format(strain, date)  # outputFolder
    if check == 2:
        folderOut = folderOut + 'x'
    else:
        folderOut = folderOut
    j = 1
    folder = folderOut + 'Emb{0}/'.format(j)
    while os.path.exists(folder):
        fileName = glob.glob(folder + '*_T01_Z01_C1.tif')
        if len(fileName) > 0:
            fileName = fileName[0].split('/')[-1]
            if fileName.split('_')[2] == date:
                if i == 1:
                    break
                else:
                    i -= 1
            j += 1
            folder = folderOut + 'Emb{0}/'.format(j)
        else:
            j += 1
            folder = folderOut + 'Emb{0}/'.format(j)
    fileName = '{0}_Emb{1}_{2}_W{3:0>2}F{4}_'.format(ri, j, date, well + 1, point + 1)

    ''' correct attenuation and save local '''
    if not os.path.exists(folder):
        os.makedirs(folder)
    else:
        print('file exist, clearing folder', folder)
        clearFolder(folder)
        if os.path.exists(folder + 'batch-output'):
            shutil.rmtree(folder + 'batch-output')
    saveImgs(im1, folder, fileName, 1)
    saveImgs(im2, folder, fileName, 2)
    saveImgs(im3, folder, fileName, 3)

    ''' populate aspect ratio file '''
    print('saveEmb aspect j=', j)
    addAspect(r, date, ri, j)


def saveImgs(imgs, folder, fileName, c):
    imgs = np.reshape(imgs, (-1, imgs.shape[-2], imgs.shape[-1])).astype(np.uint16)
    for i in range(imgs.shape[0]):
        cv2.imwrite(folder + fileName + 'T{0:0>2}_Z{1:0>2}_C{2}.tif'.format(i // z + 1, i % z + 1, c), imgs[i])


def addAspect(r, date, ri, j):
    '''
    Adds aspect ratio of an embryo into the csv file
    
    Parameters:
    r: aspect ratio
    date: date
    ri: RNAi conditions
    j: embryo number
    '''

    import operator
    newData = []
    oldData = loadAspects(aspectRatioFile)
    i = 0
    while i <= len(oldData):
        if oldData[i][0] != ri:
            newData.append(oldData[i])
        elif oldData[i][1] != date:
            newData.append(oldData[i])
        elif int(oldData[i][2]) < j:
            newData.append(oldData[i])
        i += 1
    newData.append([ri, date, '{0:0>3}'.format(j), str(r)])
    newData = sorted(newData, key=operator.itemgetter(1, 2, 3))
    saveAspects(aspectRatioFile, newData)


def loadAspects(fileName):
    '''
    reads aspect ratios from file. The aspect ratios are sorted by rnai condition, date, embryo number.
    fileName: name of the file to read from
    '''
    fileData = []
    try:
        csvFile = csv.reader(open(fileName, 'rU'), delimiter=',')  # universal
        for row in csvFile:
            fileData.append(row)
    except:
        pass
    return fileData


def saveAspects(fileName, data):
    with open(fileName, 'w') as f:
        writer = csv.writer(f, delimiter=',')
        writer.writerows(data)


if __name__ == '__main__':
    getAllEmb(folderIn)
    print('ALL DONE!!!!! :)')