Characteristics and Technical Principles of Digital Imaging

Comparison of Analog and Digital Imaging Modalities

There are two primary categories of images utilized in medical imaging: analog and digital. Analog images represent the format that humans visually interpret. This category includes common items such as standard photographs, artistic paintings, and television imagery, as well as medical images that are traditionally recorded on physical film or shown on various display devices, such as computer monitors. The information perceived in an analog image consists of varying levels of color and brightness, which is otherwise referred to as film density. A defining characteristic of analog images is that they are generally continuous and are not segmented into numerous small individual sections.

In contrast, digital images are stored and recorded as a collection of numerical data. These images are structured as a matrix or a spatial array composed of small components known as picture elements, or pixels. Every individual pixel in the array is assigned a specific numerical value. The significant advantage of the digital format is that it allows the image to be processed and manipulated in various ways by advanced computer systems.

Structural Components of Digital Imaging Systems

Within imaging and computer systems, a digital image is represented by numbers organized as binary digits, which are commonly called bits. The progression of image formation involves first dividing the image into a matrix of pixels, followed by the representation of each pixel through a dedicated series of bits. The total number of pixels contained within an image is a function of the matrix size. The specific number of bits allocated to each pixel is known as the pixel bit depth.

Technically, a digital image is a matrix of pixels where each pixel, or picture element, possesses a numerical value that corresponds to a specific brightness or color value designated for computer system interpretation. For example, a digital image matrix might contain specific numerical sequences such as $32, 32, 34, 36, 38, 42, 58, 84$ in the first row, followed by rows containing sequences like $26, 26, 26, 27, 65, 68, 76, 84$ and $95, 95, 95, 94, 94, 94, 93, 93$. Other typical values found in such a matrix include $32, 32, 34, 36, 38, 42, 66, 84$, then $26, 26, 26, 27, 65, 68, 76, 84$, followed by $28, 28, 29, 30, 52, 60, 83, 84$, and finally rows like $32, 32, 34, 36, 38, 42, 56, 84$ and $26, 26, 28, 27, 65, 68, 76, 84$.

Grayscale Bit Depth and Intensity

For images displayed in grayscale, the bit depth is a crucial metric that quantifies the total number of unique shades of gray available for display. Examples of these bit depths include $1$-bit, $2$-bit, $4$-bit, $6$-bit, $8$-bit, and $10$-bit configurations. Higher bit depths allow for a more nuanced representation of image intensity.

Field of View (FOV) and Geometric Relationships

The term Field of View ($\text{FOV}$) is used synonymously with the x-ray field dimensions. It defines the specific amount of the patient's body part or anatomical area included within the image. A direct relationship exists where a larger $\text{FOV}$ equates to a larger physical area being imaged. Crucially, changes in the $\text{FOV}$ will not change the overall matrix size; however, any change made to the matrix size will directly impact the size of the individual pixels.

Contrast Resolution and Dynamic Range

In the field of radiology, contrast resolution refers to the specific capability of an imaging modality to detect and distinguish between different levels of image intensity. The inherent level of contrast resolution in a digital image is dictated by the total count of possible pixel values. This is defined by the number of bits per pixel value, also known as the grayscale bit depth. The primary metric used to describe contrast resolution is dynamic range, which is the span of tonal difference between the most intense light and the deepest dark areas within an image. Higher dynamic range suggests the potential for more shades to be represented, though it does not guarantee a direct correlation with the exact number of tones actually reproduced in the final output.

Brightness and Spatial Resolution Principles

Brightness is a display quality that is not dictated by the technical factors selected during the initial acquisition of the image. Instead, brightness is managed by the computer hardware and can be adjusted by the user through window leveling. In contrast, spatial resolution or spatial frequency refers to the quantity of details that can be fit into a specific physical space. To determine the spatial frequency of a system, one must first calculate the pixel size using the following formula:

$\text{Pixel Size} = \frac{\text{Field of View}}{\text{Matrix Size}}$

As a practical calculation, if the field of view is $30\,\text{cm}$ and the matrix size is $1024$, the pixel size is determined as:

$\frac{30\,\text{cm}}{1024} = 0.029\,\text{cm}$

This is equivalent to $0.29\,\text{mm}$. Following this, the spatial frequency can be determined using the formula:

$\text{Spatial Frequency} = \frac{1}{2 \times \text{Pixel Size}}$

Using the previous example:

$\frac{1}{2 \times 0.29\,\text{mm}} = \frac{1}{0.58\,\text{mm}} = 1.72\,\text{lp/mm}$

Spatial Frequency and Line Pairs

Spatial frequency is expressed in units of line pairs per millimeter ($\text{lp/mm}$). There is a direct correlation between these factors: the higher the spatial frequency, the higher the spatial resolution of the image. In digital imaging environments, the ultimate limit of spatial resolution is defined by the size of the pixels.

Understanding Noise and Signal-to-Noise Ratio (SNR)

Noise is defined as any factor that interferes with image acquisition and Subsequently degrades image quality. There are several specific types of noise. Quantum noise, also referred to as quantum mottle or photon starvation, presents as a mottled appearance. This occurs when an insufficient number of photons strike the image receptor. This can be mitigated by increasing the x-ray signal through higher milliampere-seconds ($mAs$) or peak kilovoltage ($kVp$). System noise consists of random data generated by the electronic components of the imaging equipment. Ambient noise is random information caused by background radiation that strikes the receptor. Finally, scatter is noise originating from the patient themselves.

Signal-to-Noise Ratio ($\text{SNR}$) measures the relationship between meaningful signal and unwanted noise. A higher $\text{SNR}$ means there is more signal than noise. To improve $\text{SNR}$, the number of photons striking the receptor must be increased by adjusting $mAs$, $kVp$, or both. A lower $\text{SNR}$ leads to more noise and less signal, which reduces image contrast and overall quality. While maintenance can reduce system noise and grids/$kVp$ selection can reduce scatter, ambient noise is pervasive and cannot be fully eliminated as background radiation is always present.

Exposure Latitude and Diagnostic Quality

Exposure latitude in radiography identifies the range of exposure values ($mAs$ and $kVp$) within which a clinical image can be produced with acceptable diagnostic quality. Maintaining a healthy exposure latitude is vital because it provides flexibility in settings without ruining image quality. A wider exposure latitude ensures that minor variations, such as slight overexposure or underexposure, do not result in images that are clinically unusable.

Detective Quantum Efficiency (DQE) and Modulation Transfer Function (MTF)

Detective Quantum Efficiency ($\text{DQE}$) measures the efficiency of an imaging system in converting the remnant x-ray beam into a high-quality image. Systems with higher $\text{DQE}$ ratings are more efficient, meaning they can use a lower x-ray signal (lower patient dose) to produce high-quality images with better contrast and less noise. Conversely, a lower $\text{DQE}$ system requires more radiation to achieve equivalent quality. Cassette-less systems generally have higher $\text{DQE}$ ratings than cassette-based systems, and systems with no light conversion have the highest ratings of all. A perfect score of $100\%$ $\text{DQE}$ implies that entire x-ray signal is captured and converted with no added noise, which is theoretically impossible today.

The Modulation Transfer Function ($\text{MTF}$) describes the system's ability to record available spatial frequencies. It quantifies how much each individual component of the system contributes to the overall efficiency. It is defined as a ratio of the resulting image to the original object. A perfect imaging system would yield an $\text{MTF}$ of $1$, or $100\%$.

Factors Affecting Exposure Indicators

Exposure indicators are highly sensitive to several operational factors. Key among these is collimation, specifically when the system fails to correctly identify the collimated borders of the field. Other significant factors include the placement of gonadal shielding within the exposed x-ray field and the presence of medical implants or prostheses within the patient.

Image Receptor Technologies and Capture Methods

There are several options for image receptors, including Photostimulable Phosphor ($\text{PSP}$) systems, which are cassette-based, and Flat Panel Detectors ($\text{FPD}$), which are cassette-less. Digital radiography capture is divided into two methods. Indirect Capture involves absorbing x-rays, converting those x-rays into light, converting that light into an electrical signal, and then sending that signal to an Analog-to-Digital Converter ($\text{ADC}$). Direct Capture involves absorbing x-rays and converting them directly into an electrical signal, which is then sent to the $\text{ADC}$.

Specifics of Photostimulable Phosphors (PSP) and Flat Panel Detectors

$\text{PSP}$ storage plates are composed of a polyester base with a phosphor layer on one side made of europium-activated barium fluorohalide. After exposure, these plates are scanned in a $\text{PSP}$ reader before being displayed. Flat Panel Detectors use an x-ray absorber material coupled to a Thin Film Transistor ($\text{TFT}$), a Charge Coupled Device ($\text{CCD}$), or a Complementary Metal-Oxide Semiconductor ($\text{CMOS}$). The $\text{TFT}$ acts as a base for the image receptor in both direct and indirect capture modes.

CCD and CMOS Sensor Technologies

A Charge Coupled Device ($\text{CCD}$) requires a scintillator to convert x-rays into visible light. This light is then converted into an electrical charge before being read out by an external circuit. $\text{CCD}$ technology is known for high image resolution, large dynamic range, and excellent quality, although it is often more expensive and requires complex readout processes.

Complementary Metal-Oxide Semiconductors ($\text{CMOS}$) convert x-ray photons into electrical signals through an array of photodiodes and amplifiers. In a $\text{CMOS}$ sensor, every pixel is directly connected to the output through independent amplifiers, allowing charges to be transmitted immediately. The benefits of $\text{CMOS}$ include lower power consumption, faster image readout/processing, improved noise reduction, and a lower overall cost.