Interior image finder.pl
Jump to navigation
Jump to search
#!/usr/bin/perl
# Assumes that you already went through and cleaned out the completely
# blank slides
# The point list csv and the images have to be in the same directory.
# This script operates on a directory of Zeiss created microscope images
# and a Zeiss point list describing their relative positions, and returns a list
# of images that are the outermost edge images.
# The purpose of this script is to be able to derive with images are the outermost
# to be able to exclude them from wndchrm processing.
# The format of the .csv file read is as follows:
# "Slide"
# "Name","Width","Height","Description"
# "Slide2",76000,24000,"Slide - 76 mm x 24 mm (3 x 1)"
#
#
# "Positions"
# "Comments","PositionX","PositionY","PositionZ","Color","Classification"
# "1",26638,48717,23995.750,"blue","blue"
# "2",26638,48919,23995.750,"blue","blue"
# ...
# ...
use strict;
use warnings;
use Getopt::Long;
my $SlideNumber = "3883";
my $_16bit = 0;
my $select_number = undef;
GetOptions( "slide=s" => \$SlideNumber, # =s implies the value is a string
"16bit=s" => \$_16bit,
"n=i" => \$select_number );
my $CSVFileName = "slide" . $SlideNumber . ( $_16bit ? "_16bit" : "" ) . ".csv";
print $CSVFileName . "\n";
my $DEBUG0 = 0;
open CSVFILE, $CSVFileName or die $!;
# All information is stored in the @point_list master array.
# @point_list is an array of references to hashes with key-value pairs
# associated with XPos, YPos, ZPos, Color, FileExists, FileList, and NeighborList
my @point_list;
my %color_hash;
my $line_num = 0;
my $href;
my $total_line_count = 0;
my $total_file_count = 0;
while (my $line = <CSVFILE>)
{
$line_num ++;
if( $DEBUG0 ) {
print $line_num . ": " . $line;
}
# Read in the first seven lines and disregard, as they are comments
if( $line_num <= 7 ) {
next;
}
if( $line =~ /"(\d+)",(\d+\.?\d*),(\d+\.?\d*),(\d+\.?\d*),"(\w+)"/ ) {
if( $DEBUG0 ) {
print "matched first $1; second $2; third $3; fourth $4; fifth $5\n";
}
$total_line_count++;
$point_list[$1] = {
XPos => $2,
YPos => $3,
ZPos => $4,
Color => $5
};
if( !defined $color_hash{$5} ) {
$color_hash{$5} = 1;
} else {
$color_hash{$5}++;
}
}
}
if( $DEBUG0 ) {
for $href (@point_list) {
print "{";
for my $key ( keys %$href ) {
print "$key=$href->{$key} ";
}
print "}\n";
}
}
# Read in the tiffs from the current directory and cross-reference that list against
# csv list we just imported. Usually, tiffs that contain completely a blank image are
# deleted immediately after the imaging run. Therefore, There may not be an actual
# file for each line in the point list.
my $DEBUG1 = 0;
my $prefix = "monkey_muscle_" . ($_16bit ? "16bit_" : "" ) . $SlideNumber;
if( $DEBUG1 ) {
print "Looking for files that match the prefix $prefix...\n";
}
my @files = <*>;
foreach my $filename (@files) {
# files can be in the following formats:
# H_monkey_muscle_3883_92.tif
# monkey_muscle_3883_104.tif
# E_monkey_muscle_3883_57.tif
# Also rarely :
# monkey_muscle_171A_0001_69.tif
# a case which we'll ignore for now
if( $filename =~ /${prefix}_(\d+)/ ) {
if( $DEBUG1 ) {
print "\tFile $filename corresponts with Image $1.\n";
}
if( !defined $point_list[$1]->{FileExists} ) {
$point_list[$1]->{FileExists} = 1;
# If the deconvolved images exist in the directory side by side
# just count them as one.
$total_file_count++;
}
push @{ $point_list[$1]->{FileList} }, $filename;
}
}
# A non-sequential array
# will be filled out in loop below.
my %file_doesnt_exist_list;
my $DEBUG2 = 0;
if( $DEBUG2 ) {
print "Begin iterating over the colors: " . join(", ", %color_hash) . "\n";
}
# A critical piece of information is the delta x and delta y values that the
# microscope moves to go to the next image.
my $deltaX = 282;
my $deltaY = 202;
# my ($maxX, $maxY);
my ($XDist, $YDist);
my ($i, $j);
my ($this_image, $that_image);
# Iterate over each color, because each color represents a contiguous
# slab of specimen on the slide
for my $color ( keys %color_hash ) {
if( $DEBUG2 ) {
print "\tFor Color: " . $color . "\n";
}
for( $i = 0; $i < $#point_list; $i++) {
if( !defined $point_list[$i] ) {
next;
}
$this_image = $point_list[$i];
if( $this_image->{Color} ne $color ) {
if( $DEBUG2 ) {
print "\t\tImage $i is not $color, but is $this_image->{Color}. Skipping for now...\n";
}
next;
}
if( !defined( $this_image->{FileExists} ) ) {
if( $DEBUG2 ) {
print "\t\tImage $i does not have files associated with it. Skipping...\n";
}
$file_doesnt_exist_list{$i} = 1;
next;
}
if( $DEBUG2 ) {
print "\t\tImage $i is $color. Proceeding with proximity analysis.\n";
}
for( $j = 0; $j < $#point_list; $j++ ) {
if( !defined $point_list[$j] ) {
next;
}
$that_image = $point_list[$j];
if( $this_image == $that_image ) {
if( $DEBUG2 ) {
print "\t\t\tWon't compare Image $i to itself.\n";
}
next;
}
if( $DEBUG2 ) {
print "\t\t\tComparing images $i and $j.\n";
}
# Assumes all coordinates are positive
$XDist = abs( $this_image->{XPos} - $that_image->{XPos} );
$YDist = abs( $this_image->{YPos} - $that_image->{YPos} );
if( $DEBUG2 ) {
print "\t\t\t\tX Dist: $XDist, Y Dist: $YDist\n";
}
if( $XDist <= $deltaX && $YDist <= $deltaY) {
push @{ $this_image->{NeighborList} }, $j;
}
}
if( $DEBUG2 ) {
print "\t\t\tNeighbor List for Image $i: " .
join( " & ", @{ $this_image->{NeighborList} } ) . "\n";
} } # end iterating over @point_list
} # end iterating over %color_hash
my $DEBUG3 = 0;
if( $DEBUG3 ) {
print "First round of image neighbor analysis:\n";
}
for( $i = 0; $i < $#point_list; $i++ ) {
if( defined $point_list[$i]->{NeighborList} ) {
if( $DEBUG3 ) {
print "\tImage $i neighbors these images " . join( " & ", @{ $point_list[$i]->{NeighborList} } ) . "\n";
print "Number of neighbors Image $i has: " . (1 + $#{ $point_list[$i]->{NeighborList} } ) . "\n" ;
}
if( (1 + $#{ $point_list[$i]->{NeighborList} } ) == 8 ) {
$point_list[$i]->{Border} = 1;
}
if( (1 + $#{ $point_list[$i]->{NeighborList} } ) < 8) {
$point_list[$i]->{Edge} = 1;
}
}
}
my $DEBUG4 = 0;
my $round2 = 1;
my @file_doesnt_exist_array = sort keys %file_doesnt_exist_list;
my @border_images;
my @interior_images;
my @exterior_images;
if( ! $round2 ) {
# If there isn't a second round of edge analysis, just go straight to results
for( $i = 0; $i < $#point_list; $i++ ) {
push( @interior_images, $i) if( defined $point_list[$i]->{Border} );
push( @exterior_images, $i) if( defined $point_list[$i]->{Edge} );
}
}
else {
if( $DEBUG4 ) {
print "SECOND ROUND OF INTERIOR ANALYSIS...\n";
}
# FOR THIS ROUND, WE ONLY ITERATE OVER THOSE IMAGES LABELLED "BORDER"
# IN THE LAST ROUND. WHEN WE FIGURE OUT INTERIOR IMAGES OUT OF THIS REDUCED POOL
# WE'LL ADD THE ADDITIONAL LABEL "INTERIOR"
# These vars were declared above.
# my $deltaX = 282;
# my $deltaY = 202;
# my ($maxX, $maxY);
# my ($XDist, $YDist);
# my ($i, $j);
# my ($this_image, $that_image);
# Iterate over each color, because each color represents a contiguous
# slab of specimen on the slide
for my $color ( keys %color_hash ) {
if( $DEBUG4 ) {
print "\tFor Color: " . $color . "\n";
}
for( $i = 0; $i < $#point_list; $i++) {
if( !defined $point_list[$i] || !defined $point_list[$i]->{Color} ) {
next;
}
$this_image = $point_list[$i];
if( $this_image->{Color} ne $color ) {
if( $DEBUG4 ) {
print "\t\tImage $i is not $color, but is $this_image->{Color}. Skipping for now...\n";
}
next;
}
if( !defined( $this_image->{FileExists} ) ) {
if( $DEBUG4 ) {
print "\t\tImage $i does not have files associated with it. Skipping...\n";
}
$file_doesnt_exist_list{$i} = 1;
next;
}
if( defined( $this_image->{Edge} ) ) {
if( $DEBUG4 ) {
print "\t\tImage $i was determined last round to be an edge image. Skipping...\n";
}
next;
}
if( $DEBUG4 ) {
print "\t\tImage $i is $color and is a not an edge image. Proceeding with proximity analysis.\n";
}
for( $j = 0; $j < $#point_list; $j++ ) {
if( !defined $point_list[$j] || !defined $point_list[$j]->{XPos} ) {
next;
}
if( defined $point_list[$j]->{Edge} ) {
next;
}
$that_image = $point_list[$j];
if( $this_image == $that_image ) {
if( $DEBUG4 ) {
print "\t\t\tWon't compare Image $i to itself.\n";
}
next;
}
if( $DEBUG4 ) {
print "\t\t\tComparing images $i and $j.\n";
}
# Assumes all coordinates are positive
$XDist = abs( $this_image->{XPos} - $that_image->{XPos} );
$YDist = abs( $this_image->{YPos} - $that_image->{YPos} );
if( $DEBUG4 ) {
print "\t\t\t\tX Dist: $XDist, Y Dist: $YDist\n";
}
if( $XDist <= $deltaX && $YDist <= $deltaY) {
push @{ $this_image->{InteriorNeighborList} }, $j;
}
}
if( $DEBUG4 ) {
if( defined $this_image->{InteriorNeighborList} ) {
print "\t\t\tNeighbor List for Image $i: " .
join( " & ", @{ $this_image->{InteriorNeighborList} } ) . "\n";
}
}
} # end iterating over @point_list
} # end iterating over %color_hash
my $DEBUG5 = 0;
if( $DEBUG5 ) {
print "Second round of image neighbor analysis:\n";
}
for( $i = 0; $i < $#point_list; $i++ ) {
if( defined $point_list[$i]->{InteriorNeighborList} ) {
if( $DEBUG5 ) {
print "\tImage $i neighbors these images " . join( " & ", @{ $point_list[$i]->{InteriorNeighborList} } ) . "\n";
print "\tNumber of neighbors Image $i has: " . (1 + $#{ $point_list[$i]->{InteriorNeighborList} } ) . "\n" ;
}
if( (1 + $#{ $point_list[$i]->{InteriorNeighborList} } ) == 8 ) {
$point_list[$i]->{Interior} = 1;
$point_list[$i]->{Border} = undef;
}
}
}
# If there isn't a second round of edge analysis, just go straight to results
for( $i = 0; $i < $#point_list; $i++ ) {
push( @interior_images, $i) if( defined $point_list[$i]->{Interior} );
push( @border_images, $i) if( defined $point_list[$i]->{Border} );
push( @exterior_images, $i) if( defined $point_list[$i]->{Edge} );
}
}
my $DEBUG6 = 0;
if( $DEBUG6 ) {
print "The following images do not have a file associated with them:\n";
foreach ( @file_doesnt_exist_array ) {
print $_ . ", ";
}
print "\n\n";
print "List of exterior images: \n";
foreach ( sort @exterior_images ) {
print $_ . ", ";
}
print "\n\n";
if( $round2 ) {
print "List of border images: \n";
foreach ( sort @border_images ) {
print $_ . ", ";
}
print "\n\n";
}
print "List of interior images: \n";
foreach ( sort @interior_images ) {
print $_ . ", ";
}
print "\n\n";
}
my $DEBUG7 = 0;
if( $DEBUG7 ) {
print "Statistics: \n";
print "Total images mentioned in csv: $total_line_count \n";
print "Total number of corresponding files: $total_file_count \n";
print "Total number of images without files: " . ($#file_doesnt_exist_array + 1) . "\n";
print "Total number of exterior images: " . ($#exterior_images + 1) ."\n";
print( "Total number of border images: " . ($#border_images + 1) ."\n") if( $round2 );
print "Total number of interior images: " . ($#interior_images + 1) ."\n";
}
if( defined $select_number ) {
my $DEBUG8 = 0;
if( $#interior_images < 0 ) {
# If there are no interior images, go to the border images
@interior_images = @border_images;
}
if( $select_number > ( $#interior_images +1 ) ) {
# die "Unable to select $select_number images. Only $#interior_images interior images exist.\n";
$select_number = $#interior_images;
}
if( $DEBUG8 ) {
print "Original order: \n";
for ($i = 0; $i < $#interior_images; $i++ ) {
print "($i -> $interior_images[$i]) ";
}
print "\n\n";
}
for ($i = @interior_images; --$i; ) {
my $j = int rand ($i+1);
# print "i: $i j: $j \n";
next if $i == $j;
@interior_images[$i,$j] = @interior_images[$j,$i];
}
if( $DEBUG8 ) {
print "New order: \n";
for ($i = 0; $i < $#interior_images; $i++ ) {
print "($i -> $interior_images[$i]) ";
}
print "\n\n";
}
# Output a list of files associated with the first $select_number images
# in the randomized array
for( $i = 0; $i < $select_number; $i++ ) {
if( defined $point_list[$interior_images[$i]]->{FileList} ) {
foreach (@{ $point_list[$interior_images[$i]]->{FileList} } ) {
print $_ . "\n" ;
}
}
}
}
#1;
