BIO403_Lecture6_image_manipulation_2025.pdf

# Data representation and its misrepresentation This topic explores how data can be altered during its representation, leading to potential inaccuracies, and defines key forms of research misconduct [1](#page=1) [2](#page=2) [3](#page=3). ### 1.1 Understanding data representation Data representation refers to the process where information is captured and recorded. However, during this process, several issues can arise [1](#page=1) [2](#page=2): * **Loss of data:** Information can be omitted or not captured [1](#page=1) [2](#page=2). * **Addition of data:** Extraneous or incorrect information may be introduced [1](#page=1) [2](#page=2). * **Change of data:** Original data can be altered [1](#page=1) [2](#page=2). * **Manipulation of data:** Data can be intentionally altered to create a misleading impression [1](#page=1) [2](#page=2). It is crucial to recognize that data itself is not the same as its representation. The way data is presented can significantly distort its original meaning or accuracy [2](#page=2). > **Tip:** Think of data representation like making a photocopy of a document. The photocopy is a representation, but it might be darker, lighter, have smudges, or miss certain details compared to the original document. The representation is not the original itself [1](#page=1). ### 1.2 Research misconduct definitions The Department of Health and Human Services defines research misconduct as the fabrication, falsification, or plagiarism that occurs when proposing, performing, or reviewing research results [3](#page=3) [4](#page=4). #### 1.2.1 Fabrication Fabrication involves creating entirely made-up data or results and then recording or reporting them as if they were genuine [4](#page=4). #### 1.2.2 Falsification Falsification entails manipulating research materials, equipment, or processes. It also includes the act of changing or omitting results in a way that leads to the research record not accurately reflecting the true findings [4](#page=4). #### 1.2.3 Plagiarism Plagiarism is defined as the unauthorized use of another person's ideas, processes, results, or words without giving them proper acknowledgment. This can manifest in various forms, sometimes referred to as "citation amnesia," "disregard syndrome," or "bibliographic negligence." [4](#page=4). --- # Scientific image manipulation: examples and consequences Scientific image manipulation refers to the alteration of digital images used in research, which can range from acceptable enhancement to serious misconduct, leading to significant consequences. Digital images are inherently easy to manipulate due to their digital nature [15](#page=15) [5](#page=5) [6](#page=6) [7](#page=7) [9](#page=9). ### 2.1 Types of image manipulation and their implications Image manipulation can create false impressions and mislead the scientific community and the public. While some adjustments may be made to improve clarity, others can obscure or fabricate data. The core ethical issue lies in whether the manipulation alters the scientific integrity of the data [10](#page=10) [11](#page=11) [12](#page=12) [13](#page=13) [14](#page=14) [24](#page=24) [9](#page=9). #### 2.1.1 Removal and addition of data A common form of manipulation involves the selective removal or addition of parts of an image. This can include [24](#page=24) [25](#page=25): * **Selective removal of parts:** Removing specific features or data points that do not support a hypothesis or that are considered artifacts [24](#page=24) [25](#page=25). * **Selective addition of parts:** Introducing elements or data that were not originally present in the image [25](#page=25). > **Tip:** The deliberate removal or addition of data fundamentally distorts the original findings and is considered scientific misconduct. #### 2.1.2 Inappropriate replication of data Another problematic form of manipulation is the inappropriate replication of data, where elements from one part of an image are duplicated to represent data in another part, creating the illusion of multiple, independent experimental results. This is particularly egregious when it involves 'loading controls' in experiments such as Western blots [26](#page=26). > **Example:** In a Western blot analysis, if the same band pattern for a loading control is presented for two different experiments (Experiment A and Experiment B) when they should be distinct, it suggests the data has been fabricated or manipulated to appear consistent [26](#page=26). #### 2.1.3 Selective adjustment of image parts Manipulation can also involve selectively adjusting specific parts of an image, such as altering contrast or brightness in particular regions to highlight certain features or obscure others. While global adjustments might be acceptable for improving overall image quality, localized adjustments can disproportionately affect specific data points and lead to misinterpretation [29](#page=29). #### 2.1.4 Exaggerated global adjustments Exaggerated global adjustments, such as drastically altering contrast or brightness across the entire image, can also distort the original data. This might make faint signals appear stronger or suppress background noise excessively, leading to an inaccurate representation of the experimental results [31](#page=31). #### 2.1.5 Ambiguity in image selection Sometimes, the manipulation lies in the selection of which image or part of an image to present, offering multiple options that may not accurately reflect the full data set [32](#page=32). ### 2.2 Examples of scientific image manipulation in publications Several published cases highlight instances of scientific image manipulation: * **Maile et al., Science, 2004** [16](#page=16) [17](#page=17). * **Beisel et al., Nature, 2002** [16](#page=16) [17](#page=17). * **Sanchez-Elsner et al., Science, 2006**: This case involved issues with image similarity and potential manipulation of gamma settings and aspect ratios (V/H stretching) [18](#page=18) [19](#page=19) [20](#page=20) [21](#page=21) [22](#page=22) [23](#page=23). * **Sud N et al., Am J Physiol Lung Cell Mol Physiol, 2008**: This study presented cases of inappropriate replication of data across different figures and panels [27](#page=27) [28](#page=28). * **Uittenbogaard A, et al., J Biol Chem, 2002** [38](#page=38) [39](#page=39) [40](#page=40). * **Sawada M, et al., Nat Cell Biol, 2003** [41](#page=41) [42](#page=42) [43](#page=43). * **Hwang WS, et al., Science, 2005** [44](#page=44). * **Kang J et al., Cell, 2005**: This involved enhanced contrast in original images, potentially obscuring details or making weak signals more prominent [45](#page=45) [46](#page=46) [47](#page=47). ### 2.3 Consequences of scientific image manipulation The consequences of scientific image manipulation can be severe, undermining the integrity of research and potentially leading to: * **Retractions of publications:** Journals may retract papers containing falsified or manipulated images [41](#page=41) [45](#page=45). * **Damage to scientific reputation:** Researchers involved in image manipulation face significant damage to their credibility and career prospects [37](#page=37). * **Loss of public trust:** Scientific misconduct involving image manipulation erodes public trust in science and its findings. * **Misguided future research:** Manipulated data can lead other researchers down incorrect paths, wasting resources and time. ### 2.4 Acceptable image manipulation practices Not all image manipulation is unethical. Certain practices are considered acceptable when they are used to enhance data clarity and are properly disclosed. These include: * **Adding arrows:** Using arrows to point out specific features of interest in an image [50](#page=50) [51](#page=51). * **Pseudocoloring particles:** Applying pseudocolors to highlight features like immunogold particles without altering the brightness of individual pixels [50](#page=50) [51](#page=51). It is crucial that any such enhancements are disclosed in the figure legend [50](#page=50) [51](#page=51). > **Note:** The key distinction between acceptable and unacceptable manipulation lies in whether the alterations distort the original data or introduce false information. Transparency and disclosure are paramount [50](#page=50) [51](#page=51). --- # Guidelines and prevention of scientific image manipulation This section outlines journal guidelines for acceptable scientific image manipulation and details preventive measures and checks implemented by publishers to ensure data integrity. ### 3.1 Publisher scrutiny and prevention measures All digital images submitted for publication are scrutinized by the production department for any indications of improper manipulation. If questions arise, the production department refers them to the editors, who then request the original data from authors for comparison with the prepared figures. Failure to produce original data can lead to the revocation of manuscript acceptance. Deliberate misrepresentation of data will result in acceptance revocation and reporting to the author's institution or funding agency [52](#page=52). > **Tip:** Publishers employ image checking systems, such as those used by JCB, JEM, and JGP (Rockefeller University Press), to detect image manipulation [52](#page=52) [54](#page=54). ### 3.2 General guidelines for image manipulation #### 3.2.1 Acceptable manipulations * Adjustments of brightness, contrast, or color balance are acceptable if they are applied to the *entire* image and do not obscure or eliminate any information present in the original [53](#page=53). * Cropping an image is generally considered acceptable. However, cropped gels in a paper must retain important bands and cropped blots should retain at least six band widths above and below the band of interest [56](#page=56). * Simple adjustments to the entire image are usually acceptable . * Manipulation of images should only be performed on a copy of the unprocessed image . #### 3.2.2 Unacceptable manipulations * No specific feature within an image may be enhanced, obscured, moved, removed, or introduced [53](#page=53). * The use of touch-up tools, such as cloning and healing tools in software like Photoshop, or any feature that deliberately obscures manipulation, is to be avoided [55](#page=55). * Contrast should not be adjusted to the point where data disappear [55](#page=55). * Threshold manipulation, expansion or contraction of single ranges, and altering high signals should be avoided [57](#page=57). * Manipulations performed on one area of an image but not on other areas are questionable . * Cloning or copying objects from other parts of the same image or from a different image is very questionable . * Use of software filters to improve quality is not recommended for biological images . * Avoid the use of lossy compression . #### 3.2.3 Disclosure requirements * Nonlinear adjustments, such as changes to gamma settings, must be disclosed. Pseudo-coloring and nonlinear adjustments are only allowed if unavoidable and must be disclosed [53](#page=53) [57](#page=57). * The grouping of images from different parts of the same gel, or from different gels, fields, or exposures, must be made explicit by the figure's arrangement (e.g., using dividing lines) and in the figure legend [53](#page=53). * If juxtaposing images is essential, the borders should be clearly demarcated in the figure and described in the legend [54](#page=54). * When submitting revised final figures, authors may be asked to submit original, unprocessed images [55](#page=55). * Authors should list all image acquisition tools and image software used [54](#page=54). * Key image-acquisition settings and processing manipulations should be documented in the Supplementary Information [54](#page=54). * Adjustments of individual color channels on "merged" images should be noted in the figure legend [57](#page=57). ### 3.3 Guidelines for gels and blots * Vertically sliced gels that juxtapose lanes not contiguous in the experiment must have a clear separation or a black line delineating the boundary [56](#page=56). * Cropped gels in the paper must retain important bands [56](#page=56). * Cropped blots in the body of the paper should retain at least six band widths above and below the band [56](#page=56). * High-contrast gels and blots are discouraged. Multiple exposures should be presented in supplementary information if high contrast is unavoidable [56](#page=56). * Immunoblots should be surrounded by a black line to indicate borders if the background is faint [56](#page=56). ### 3.4 Quantitative imaging and data integrity * For quantitative comparison, appropriate reagents, controls, and imaging methods with linear signal ranges should be used [57](#page=57). * Adjustments should be applied to the entire image [57](#page=57). * Intensity measurements should be performed on uniformly processed image data, and the data should be calibrated to a known standard . ### 3.5 Image checking systems and software Publishers like Rockefeller University Press utilize image checking systems to verify data integrity. Authors are required to list all image acquisition tools and image software used [52](#page=52) [54](#page=54). ### 3.6 Understanding pixel images Scientific images are fundamentally data that can be compromised by inappropriate manipulations [66](#page=66). #### 3.6.1 Pixel images and resolution * Pixel images are composed of a grid of pixels, where each pixel has a specific position (x, y) and intensity value. The total number of pixels ($x \times y$) determines the resolution of the image [68](#page=68). * Different pixel counts and pixel sizes affect image clarity and resolution. A larger pixel count with the same pixel size leads to better resolution [64](#page=64). #### 3.6.2 Data format and encoding * Pixel intensity values range from 0 (no light/signal) to a maximum value, often 255 for 8-bit images [74](#page=74) [75](#page=75) [76](#page=76). * Bit depth determines the number of shades of grey or colors an image can represent. For example, 8 bits allow for 256 tones, while 16 bits allow for 65,536 tones [77](#page=77) [78](#page=78). #### 3.6.3 Dynamic range and Look-Up Tables (LUTs) * Dynamic range refers to the range of intensity levels (pixel values) between the lowest and highest detectable values. It can be adjusted during acquisition (sensitivity/gain) or post-processing (brightness/contrast) [81](#page=81) [82](#page=82). * Look-Up Tables (LUTs) are used to match intensity levels to specific displayed information, affecting how image data is visualized [85](#page=85) [86](#page=86) [88](#page=88) [89](#page=89) [90](#page=90) [91](#page=91) [92](#page=92) [93](#page=93). #### 3.6.4 Saturation Saturation refers to the intensity levels of pixels, with "under-saturated" and "over-saturated" indicating extremes of the intensity range [94](#page=94). ### 3.7 Image file formats * Different image file formats (e.g., TIFF, JPEG) have varying characteristics and levels of compression [100](#page=100) . * JPEG is a lossy compression format that can degrade image quality with repeated saving or manipulation. TIFF is a lossless format . * JPEG compression involves techniques like the Discrete Cosine Transform (DCT) . ### 3.8 Non-linear adjustments Non-linear adjustments, such as gamma corrections, alter the relationship between pixel values and their displayed intensity, impacting the image's appearance. These should be disclosed when used [53](#page=53) [57](#page=57). ### 3.9 Summary of rules for scientific images 1. Scientific images are data that can be compromised by inappropriate manipulations . 2. Manipulation of images should only be performed on a copy of the unprocessed image . 3. Simple adjustments to the entire image are (usually) acceptable . 4. Cropping an image is (usually) acceptable . 5. Images that will be compared to one another should be acquired under identical conditions, and any post-acquisition image processing should also be identical . 6. Manipulations that are performed on one area of an image but not on other areas are questionable . 7. Use of software filters to improve quality is not recommended for biological images . 8. Cloning or copying objects from other parts of the same image or from a different image is very questionable . 9. Intensity measurements should be performed on uniformly processed image data, and the data should be calibrated to a known standard . 10. Avoid the use of lossy compression . 11. Magnification and resolution are important . 12. Be careful when changing the size (in Pixels) of a digital image . --- ## Common mistakes to avoid - Review all topics thoroughly before exams - Pay attention to formulas and key definitions - Practice with examples provided in each section - Don't memorize without understanding the underlying concepts

| Term | Definition | |------|------------| | Data Representation | The way in which data is structured, organized, and presented, which can influence its interpretation and may be subject to changes or manipulations. | | Research Misconduct | Defined as fabrication, falsification, or plagiarism in proposing, performing, or reviewing research results, leading to compromised scientific integrity. | | Fabrication | The act of making up data or results and then recording or reporting them as if they were genuine, which is a form of research misconduct. | | Falsification | The manipulation of research materials, equipment, or processes, or the alteration or omission of results, such that the research is not accurately represented in the record. | | Plagiarism | The appropriation of another person's ideas, processes, results, or words without giving proper credit, which includes forms like citation amnesia and bibliographic negligence. | | Digital Images | Visual data captured or created using digital technology, which are susceptible to easy manipulation due to their digital nature. | | Image Manipulation | The alteration of digital images, which can range from simple adjustments to complex modifications, and can be used both legitimately and illegitimately in scientific contexts. | | Pixel | The smallest addressable element in a raster image or the smallest controllable element of a picture on a screen; a picture element. | | Vector Graphics | Images that are composed of mathematical equations that define points, lines, and curves, allowing for infinite scalability without loss of quality. | | Raster Graphics | Images that are composed of a grid of pixels, where each pixel has a specific color and position, and scaling can lead to a loss of quality. | | Charge-Coupled Device (CCD) | An electronic sensor used in digital cameras and imagers to capture light and convert it into an electrical signal, forming the basis of pixel data. | | Resolution | The level of detail in an image, determined by the number of pixels per unit of area; higher resolution means more pixels and thus more detail. | | Bit Depth | The number of bits used to represent the color of a single pixel in a bitmap image or the number of shades of gray in a grayscale image, determining the range of tones. | | Dynamic Range | The ratio between the maximum and minimum signal values that a system can detect or display, indicating the range of light intensities that can be captured in an image. | | Look-Up Table (LUT) | A data table used to transform input values into output values, often used in image processing to adjust display characteristics like color or brightness. | | Saturation | The intensity or purity of a color in an image; high saturation means a vivid color, while low saturation means a duller color. | | JPEG | A common file format for digital images that uses lossy compression to reduce file size, often by discarding some image information. | | TIFF | A flexible image file format that supports lossless compression and is often used for high-quality images in professional contexts, preserving more image data than JPEG. | | Convolution Kernel | A small matrix used in image processing to apply filters and effects to an image by performing a convolution operation with the image pixels. | | Cloning | A digital image editing technique where pixels from one area of an image are copied and pasted onto another area, often used to conceal defects or create illusions. | | Lossy Compression | A data compression method that permanently discards some data to achieve smaller file sizes, which can lead to a loss of image quality. | | Nonlinear Adjustments | Image processing operations that do not scale linearly with the input values, such as gamma correction, which can alter the perceived brightness and contrast of an image. | | Gamma Correction | A nonlinear adjustment that modifies the luminance of an image by altering the gamma curve, affecting the overall brightness and contrast. | | Pseudocoloring | The assignment of artificial colors to grayscale images or data to enhance visualization and interpretation, especially for highlighting specific features or ranges. | | Figure Legend | A descriptive text accompanying a figure or image that explains its content, methods, and any relevant details, essential for proper interpretation. | | Image Acquisition Tools | Software and hardware used to capture or generate digital images, including cameras, microscopes, and scanners. | | Image Software | Applications used to process, edit, and manipulate digital images, such as Photoshop or GIMP. |