How do I write an original article from scratch?

Your question touches the two most difficult steps: turning a clinical question into a feasible study, and turning the study into a paper that gets published. The first is hard because it must be feasible; the second is hard because a journal must accept it. Here are the key points of the process.

To make a clinical study feasible, the most important thing is reading the literature

Reading the literature is the most important step. If your search is not thorough, the idea you came up with may be one that someone else has already done, and you will waste a lot of time. Worse, the study you design may not be feasible at all, or may not be able to answer your original clinical question. Any of these will leave your paper wandering from journal to journal without being accepted.

My method is to spend one evening searching carefully, print every paper I think I need to read, and read them during boring meetings. Don't worry: in a hospital, there is never a shortage of boring meetings. You know what I mean.

A key point: do not assume a PubMed search finds everything. Keywords often come in many versions, such as MDCT / MSCT / multi-detector CT / multi-detector row CT / multi-slice CT / multi-section CT. These all describe the same thing, and it is easy to miss one of them.

After the first PubMed search, go through the reference lists of the papers you have found to catch anything you missed. This is a practical and important technique. Read all the papers, see how others designed their studies, and think about the resources and time you have. Only then can you find the right study design and approach.

Beginners often design a study that seems obviously great to them. The common mistakes are that the study cannot actually answer the original question, and that they do not have the resources to carry it out.

Some people think: I can surely enroll 100 patients with this condition, do an MRI before and after, and publish. The problem is that each MRI is expensive, and 200 of them add up to a large sum. Whose research budget will pay for that? If you bill it to insurance, what indication will you write? If the insurer notices the unusual volume and denies every claim, who is responsible? And who in radiology will work with you, read the images, and make the measurements the study needs, in a way that the international research community will accept?

Beginners often think, "Can't I just order the scans?" No. If you only place the orders, the 200 MRI scans will be done under different conditions, and in the end you will have nothing you can write up. So reading the literature, considering all the resources available, and working out how to handle each step are very important.

Execution and details still bring uncertainty

An idea is only the beginning, and many variables follow. At this point, planning a timeline is not easy.

In my own case, after I had a first idea for the contrast-covering time concept, which I developed to solve a contrast injection problem, I went to the hospital workstation at five in the morning, pulled a few dozen past cases, and checked whether my idea was right. (The morning conference started at 7:30, and after that my time was no longer my own, so I had to go that early.)

https://www.ajronline.org/doi/full/10.2214/AJR.06.0902

Then I wrote the protocol and planned the study, submitted it to the IRB, got approval, carried it out, collected the data, checked whether the preliminary results were meaningful, made the figures and tables, and wrote with full effort.

Much of this is under your control, but IRB approval depends on when the committee meets, and enrollment depends on luck. Planning a firm timeline at the start is almost impossible. Beginners need extraordinary persistence to push a study forward. The first time will not go smoothly, but the second will be better, and after the third you will know how to handle the small office politics you can avoid in advance and the details that can go wrong, and things will go much more smoothly.

Writing and figures

Once the study is done, I suggest you first run the numbers and make a few figures to see whether they tell an interesting story. If they do, then write the paper. If the data are not good enough and you need more data or a new angle, there is still time. If you start writing the text right away and later change the direction of the analysis, everything you wrote is wasted.

In general, once the data have been run, the results are out and you are sure there is a paper, an experienced researcher can write the whole paper in about one weekend (Saturday and Sunday, 8 a.m. to 8 p.m., 24 hours in total). Beginners will be slower, but they gradually get closer to this speed. Don't worry!