READ The Read button brings up the standard MS-Windows Open-File dialog box. In addition, for the paletted mode, the palette is changed to the normal MS-Windows colors. This change is done in order to make the dialog box entries more readable. The file formats that are recognized for reading by ImprocRAD, organized by extension, are: .u08 = Unsigned 8-bit no header in 1-byte .u10 = Unsigned 10-bit no header in 2-bytes, unpacked .u12 = Unsigned 12-bit no header in 2-bytes, unpacked .u14 = Unsigned 14-bit no header in 2-bytes, unpacked .u16 = Unsigned 16-bit no header in 2-bytes .s08 = Signed 8-bit no header in 1-byte .s16 = Signed 16-bit no header in 2-bytes, unpacked .pcx = Z-soft, now Microsoft PCX format, run-length encoded, 8-bit .bmp = Microsoft bitmap, uncompressed, 8-bit. Both monochrome and color images can be read and displayed. .mat = Matlab formats, 64-bit double, 32-bit floating point, 32-bit signed integer, 16-bit signed integer, 16-bit unsigned integer, 8-bit unsigned integer .fcr = Fuji Computed Radiography format 10-bit, 2048-byte header .dcm = Dicom 3.0. A very limited subset of Dicom formats are handled. These are: - Lumisys and Vidar uncompressed unsiged - GE Tomography uncompressed signed - Toshiba Digital Fluoroscopy reversibly compressed - Philips Integris V Digital Fluoroscopy reversibly compressed - The latest DICOM standard is rev. 1999, parts 1-14. Part 14 is about the grayscale display function .jpg = JPEG reversibly compressed .tif = Uncompressed, monochrome, 8-bit and 16-bit, both Little and Big Endian byte orders. .img = Several different formats from Mike Arthur: 0: Photometrics, unsigned 12-bit, 160 Byte Header, 1016x1024 1: MedOptics unsigned 10-bit, No Header, 1024x1024 2: Infimed unsigned 12-bit, 10-byte header, 1024x1024 3: MedOptics unsigned 14-bit, 10-byte header, 1024x1024 The header consists of: - An (unsigned short) byte-order sensor that = 1 if the bytes are not swapped. - The (unsigned long) number of pixels per line - The (unsigned long) number of lines per image .ids = UCLA/LLNL Mammography Collection .tct = Toshiba CT .seq = ImprocRAD sequence of images .moi = Special format for Mike Arthur .b12 = Byte reversed 12-bit in 2bytes,no header, unpacked .mri = GE MRI images from Art Gmitro .wjd = ImprocRAD proprietary format. One kByte header containing picture dimentsions and bit-depth among other information. Accomodates bit-depths from one to 16 bits. If the extension is not recognized by ImprocRAD, either because the extension is not contained in its internal dictionary, or because the configuration is set to not sense extensions then ImprocRAD displays an extension selection box. The file is read according to the dictionary entry for the extension selected even if the actual file extension differs from that selection. If the data to be read are signed, ImprocRAD adds a bias of 32768 and then stores them internally as 16-bit unsigned data. In this way, the resulting images can be mixed with images that stem from unsigned data. If the CropOnRead flag is set, for instance through the configuration dialog box, then a CropOnRead dialog box is displayed before the image is read. The first set of entries in this box are in the ReadDimensions group. The xSize and ySize are either read from the image header, or estimated using the Autosize algorithm. The Autosize algorithm is designed for application to diagnostic radiographs. When applied to test patterns, the results are often unreliable so that the dimensions must be manually entered in the edit- control boxes. The header size is only read from files with file extensions that contain the head-size information. There is no attempt to estimate the header size of unknown file formats. There is also no attempt to estimate bit-depth. No attempt is made to estimate whether or not the bytes are swapped relative to the normal PC storage order. The PC storage order is IEEE Little Endian, the least significant byte is stored first. The Big Endian format, where the most significant byte is stored last is used for MAC's and SUN machines. The next group concerns the CroppingWindow. These parameters allow a section of the stored image to be extracted during reading so that only the smaller image is stored in memory. The parameters are the xSize and the ySize of the smaller, cropped, image. The offsets can be measured either from the center of the original image of from the origin (upper-left corner). The choice is made by pressing the corresponding button. IMPORTANT: If the offset is measured from the center of the image, then it is measured to the center of the cropping region; if the offset is measured from the origin, it is measured from the upper-left corner of the cropping region. There are two OffsetFromMouse buttons. They allow choosing the offset by positioning the mouse on an uncropped version of the image that is to be cropped. To use the mouse-cropping, first read in the image without cropping. Next, open the Cursor-Position dialog box from the 'Analyze' dialog box. Move the mouse to the desired center of the cropped image for OffsetFrom Center or the the upper-left corner of the cropped image for OffsetFromOrigin. Next, press either the MouseFromCenter or MouseFromOrigin button. The final step is to press the ReadImage button at the bottom of the CropOnRead dialog box. After reading, if desired, close the Cursor-Position dialog box. The MinifyOnRead group subsamples the image using every point, every second point, etc., up to every sixth point. In addition, the IntegerFitToScreen button calculates the subsampling interval that will allow the resulting image to be displayed on the screen in its entirity. The FloatFitToScreen performs the same function, but uses fine-grain, non-equilength interval sampling to attain a closer fit to the screen. The final dimensions are display-only, they cannot be directly edited. The ReadImage button reads the image from disk using the parameters set by applying the operations described in the preceding material. This help file was updated on January 24, 2001.