In direct conversion, x-ray photons are absorbed by the coating material and immediately converted into an electrical signal. The flat-panel detector has a radiation-conversion material or photoconductor, typically made of amorphous selenium (a-Se) that is about 500 µm thick for radiography and 200 to 250 µm thick for mammography. This material absorbs x-rays and converts them to electrons, which are stored in the TFT detectors (Figure 5-1). The TFT is a photosensitive array made up of small (about 100 to 200 µm) pixels, also called a detector element (del) in these TFT arrays. Each pixel contains a photodiode that absorbs the electrons and generates electrical charges. A field effect transistor (FET) or silicon TFT isolates each pixel element and reacts like a switch to send the electrical charges to the image processor (Figure 5-2). More than 1 million pixels can be read and converted to a composite digital image in less than 1 second. A line of TFT switches, each associated with a storage capacitor, allows the electrical charge information to discharge when the switches are closed. The information is discharged onto the data columns and read out with dedicated electronics. Specialized silicon integrated circuits are connected along the edges of the detector matrix. On one side, integrated circuits control the line scanning sequence, and on the other side, low-noise, high-sensitivity amplifiers perform the readout, amplification, and analog-to-digital conversion. High-speed digital electronics are then used to achieve fast image acquisition and processing.

Indirect-conversion detectors are similar to direct detectors in that they use TFT technology. Unlike direct conversion, indirect conversion is a two-step process: x-ray photons are converted to light, and then the light photons are converted to an electrical signal. A phosphor such as gadolinium oxysulphide (Gd₂O₂S), or thallium doped cesium iodide (CsI[TI]) rapidly absorbs x-rays and produces light. The phosphor layer is known as the scintillation layer. The scintillation layer can be either structured or unstructured. Unstructured layers produce more scattered light than structured layers, thereby decreasing the efficiency of the detector. The light is then converted into an electric charge by a photodetector such as a hydrogenated amorphous silicon (a-Si:H) photodiode array. The two-step process occurs as follows:

Gd₂O₂S Detectors.

The gadolinium oxysulphide (Gd₂O₂S) detecting material is made from small crystals bound together in an unstructured or turbid (powdered granules) layer along with a polyurethane material. The crystalline layer is filled with air pockets that allow light that was generated in the phosphor to escape. It also allows light to escape laterally before it reaches the surface, reducing the efficiency of the phosphor and lowering spatial resolution (Figure 5-4). Gd₂O₂S was used primarily for applications that required a rugged detector, such as portable detectors either wired or wireless. The ruggedness was a result of the unstructured or phosphor layer.

CsI Detectors.

The most popular type of amorphous silicon detector uses a CsI scintillator. The scintillator is made by growing very thin crystalline needles (5 to 10 µm wide) perpendicular to the detector surface; these crystalline needles work as light-directing tubes, much like fiber optics (Figure 5-5). Because of how the cesium iodide crystals are grown, they are considered a structured scintillator. These needles can be evaporated onto the surface of the layer, or they can be grown with a seed layer. This allows greater detection of x-rays (higher detective quantum efficiency DQE than a Gd₂O₂S scintillator), and because there is almost no light spread, the spatial resolution is higher (Figure 5-6). These needles absorb the x-ray photons and convert their energy into light, channeling it to the amorphous silicon photodiode array. In the past, CsI detectors could not be practically used in a portable-type detector. The crystalline structure was too delicate to be used outside of a protected environment of a fixed detector array. However, today, because of the advances in materials and crystalline formation on the array, both CsI and Gd₂O₂S are being used in portable detectors.

When the exposure is made, the sensing/storage component within the pixel contains the image information. The information is then read out line by line via the changing of the control line voltage so each pixel is connected to its corresponding data line at the bottom of the column (Figure 5-7). Figure 5-7A illustrates the detector in its initialized state, prepared and waiting for exposure. In Figure 5-7B, the detector has been exposed and is awaiting reading of the array. Figure 5-7C demonstrates the readout or release of signal of row one. Notice that a voltage is applied across the control or switch line corresponding to row one. In the final image of Figure 5-7, row one has returned to an initialized state (the process of returning to this state varies per vendor and is beyond the scope of this textbook) after release of its signal and the readout process continues to row two. Once the signal has been released, the energy is then transferred to the electronics that are attached to the edges of the array.

Gain calibration is used to correct flaws in the detector. If an area of the detector has a large amount of dead pixels or if there is an area of the detector that has poor connections between the conversion layer and the a-Si/TFT array, a noticeable artifact can be seen on the screen. To remove the potential of having that artifact interfere with diagnosis, a process known as gain calibration or flat fielding is used. This process will remove these densities on the image by creating a mask of those defects. When a subsequent image is taken, the software will use the mask to remove the unwanted densities and leave only the diagnostic information. Gain calibration should be performed under the same or similar circumstances in which clinical images will be taken. The calibration should be done according to the equipment manufacturer’s guidelines. In Figure 5-8, the automatic exposure control AEC cells appear as slightly positive squares in the image. The gain calibration was performed with the detector outside of the Bucky and then a subsequent image was taken with the detector in the Bucky. This could happen if anything is left on the detector during a gain calibration. These faint marks could lead to a misdiagnosis.

Some flat-panel systems have the ability to take images faster than the detector can accommodate. If an image is taken prior to the detector releasing all of the signal from the previous image, a faint image of the previous exposure may be visible. This is known as image lag. Image lag is almost a double exposure of sorts. When the detector is unable to get rid of all of the signal from the previous image, lag can occur. Rapid succession of images is only one reason for image lag. Another reason image lag might occur is overexposure or an area with little beam attenuation, such as a technologist’s lead marker location. In Figure 5-9, a faint R is visible in the proximal thoracic spinous processes. The R is left over from the previous image. The following techniques can be used to reduce the possibility of image lag:

Image lag can also be corrected with the detector’s software using the dark noise or offset correction. Offset correction determines the amount of signal inherent in the detector. If an offset correction is performed prior to residual signal leaving the pixels, that information will be stored as inherent and could cause a negative image of the residual signal. Always follow the equipment manufacturer’s guidelines when performing any detector calibrations. For example, in Figure 5-10 an offset correction was performed after an examination was done with a right side marker in an area of unattenuated radiation. Residual signal remained in that area and a subsequent offset correction was performed. The R is now viewed inversely because that signal created a mask during the offset correction. In Figure 5-10 it is clear that the signal is still very strong and this image was taken 10 minutes after the initial exposure.